0:08

Welcome to the Skeptics Guide

0:10

to Emergency Medicine. Meet them, greet

0:13

them, treat them, and treat them.

0:15

Today's date is February 12th 2025

0:17

and I'm your skeptical host, Dennis

0:20

Wren. The title of today's episode

0:22

is Are ESI Levels? Accurate for

0:24

triage of pediatric patients? And

0:26

our guest skeptic is Dr.

0:29

Brandon Ho. He is a graduating

0:31

pediatric emergency medicine fellow

0:33

at Children's National Hospital

0:35

in Washington DC DC.

0:38

and he is soon to be

0:40

an attending physician at Seattle Children's.

0:42

His research interests include AI

0:44

or artificial intelligence and

0:46

health care, medical education,

0:48

and social determinants of

0:50

health. Dr. Ho, welcome to S.J. Pete's.

0:53

Hi Dennis, thanks for having me

0:55

today. Very excited to be here. Well

0:57

Brandon, I'm glad that we have

0:59

an opportunity to podcast together. I

1:01

have a little bit of sadness. Tier

1:03

that you will be leaving us and

1:06

going across the country. I know, I'm

1:08

very very sad, very, I'm going to

1:10

miss you guys very much. And Brandon,

1:12

I think you also have another

1:14

new life development that is coming

1:17

along. That's very true. My wife is

1:19

due in the next couple of weeks, and

1:21

my dog is very excited about

1:23

this, or very scared. I can't really

1:26

tell. Well, congratulations to you and your wife

1:28

and your dog, I guess, because he's going

1:30

to be part of this new adventure for

1:32

you all as well. Big changes coming down

1:34

the line. Lots of big changes. Thank you,

1:36

Dennis. All right. Well, I think we

1:38

are talking about ESI levels today, which

1:40

are emergency severity index levels, and I

1:43

understand you brought us a case. So

1:45

you are approached by the medical director

1:47

of your emergency department, and she

1:49

has noticed that recently there has

1:51

been an increasing number of pediatric

1:54

cases presenting to your facility. In

1:56

some of these cases, the children end up

1:58

being more sick than initially. As

2:00

the institution's evidence-based medicine

2:03

enthusiasts, she asks you, what do

2:05

you think of the triage system we're using

2:07

now? And how accurate is it for

2:09

children? Oh, pediatric triage. It is

2:12

such a fundamental component of emergency

2:14

medicine. It is the first critical

2:16

step in how we manage acutely

2:18

ill or in your children in

2:20

the emergency department. And unlike

2:23

adult triage, pediatric triage

2:25

presents a unique set

2:27

of challenges because there's...

2:29

variations in physiology, developmental

2:31

differences, and also communication

2:33

barriers in younger patients.

2:35

Accurately assessing the severity

2:37

of a child's condition is

2:39

essential for ensuring timely intervention

2:42

while avoiding unnecessary resource utilization.

2:44

The Emergency Severity Index

2:46

is very very common and widely used

2:49

in true arch systems all over the

2:51

United States, and it classifies

2:53

patients based on acuity and

2:55

predicted resource utilization. ranging from

2:58

ESI level one, which are

3:00

generally your patients that require

3:03

immediate life-saving interventions, to ESI

3:05

level five, those that don't

3:08

need any resources beyond

3:10

the physician evaluation. And

3:12

right now pediatric triage remains

3:14

a particularly challenging problem

3:16

due to factors based

3:19

on age-based vital signs,

3:21

difficulty in obtaining accurate

3:23

histories, and... non-specific

3:25

presentations of critical

3:27

illnesses in pediatrics. And

3:29

typically, ESI levels one and two

3:32

are used to assess the acuity

3:34

or risk of instability, while ESI

3:36

levels three, four, and five are

3:38

kind of more determined by the

3:41

expected resource needs. And those resources

3:43

can be labs, imaging, medications, consultations,

3:45

stuff like that. And we'll have

3:48

a nice little table in the

3:50

show notes for you to look at. Yeah.

3:52

And, you know, as you can imagine, because

3:54

of these determinations, it

3:56

can be imperfect with previous

3:59

studies. showing mistriage rates as

4:01

high as 50%. Pediatric patients

4:04

can either be under triage, so

4:06

assuming a lower acuity level than

4:08

warranted or over triage, assigning a

4:11

higher acuity level than necessary. And

4:13

this can really have significant

4:15

consequences when emergency departments

4:18

are experiencing long wait

4:20

times, boarding patients, or

4:23

just chronically understaffed. The

4:25

problem here is that under triage.

4:28

can lead to delayed care for

4:30

critically ill children, whereas over-trias

4:32

can actually result in

4:34

unnecessary resource use, increasing

4:36

health care costs, and prolonged

4:39

ED crowding. And studies have shown

4:41

that pediatric patients are frequently

4:43

subject to both types of

4:46

errors with young children and

4:48

those presenting with at typical symptoms

4:50

being at risk. So Brandon,

4:53

what's the clinical question that we're

4:55

asking today? How accurate is ESI

4:57

version 4? in predicting acuity and

4:59

resource needs among pediatric

5:01

emergency department patients. And

5:04

what's our reference? We are looking

5:06

at a study published in

5:08

JAMA Pediatrics in October, 2024,

5:10

by SACS at all, called

5:12

Emergency Surveyity Index Version 4,

5:14

and triage of pediatric emergency

5:16

department patients. Let's break it down

5:18

now by going through our peacot questions.

5:20

What was the population that they included

5:23

in the study? They included

5:25

pediatric patients age 0 to

5:27

18 years old presenting to

5:30

21 Kaiser Permanente EDs from

5:32

January 1st 2016 to December

5:35

31st 2020. And they excluded

5:37

patients that were missing ESI

5:39

incomplete ED time variables,

5:41

transferred patients, patients who

5:43

left AMA or left

5:46

without being seen. What

5:48

was the exposure that they were

5:50

looking at? They looked at the

5:52

assigned ESI level compared

5:54

to actual resource utilization

5:57

and critical interventions. In

5:59

the comparison? The comparison was

6:01

correct triage rates against under-triage

6:03

and over-triage cases to identify

6:06

patterns of mistriage. And let's talk

6:08

about their outcome as what was the

6:10

primary outcome they were looking at? Their

6:12

primary outcome was to look at the rate

6:14

of mistriage. So under-triage

6:16

or over-triage in their pediatric

6:19

patients and their secondary outcome

6:21

were patient and visit characteristics

6:23

associated with mistriage. including things

6:26

like age, sex, probabilities, mode

6:28

of arrival, and race and

6:30

ethnicity. And finally, what type of

6:32

study was this? This was a

6:34

multi-center retrospective cohort study.

6:37

And the authors concluded,

6:39

quote, this multi-center retrospective

6:41

study found that mistriage

6:43

with ESI version 4

6:45

was common in pediatric

6:47

emergency department visits. There is

6:49

an opportunity to improve

6:52

pediatric ED triage both

6:54

in early identification of

6:56

low acuity patients with

6:58

low resource needs. Future

7:00

research should include assessments

7:03

based on version 5 of

7:05

the ESI which was released

7:08

after this study was completed.

7:10

Moving on to our quality

7:12

checklist. First question for

7:14

you Brandon. Did the

7:16

study address a clearly

7:18

focused issue? Yeah, I think

7:21

so. They were looking specifically at

7:23

ESI triage accuracy. Did

7:25

the authors use an appropriate

7:27

method to answer their question? I

7:30

think they did. Was the

7:32

cohort recruited in an acceptable

7:35

way? Yes. Was the exposure

7:37

accurately measured to minimize

7:39

bias? Yes, I think so. Was

7:41

the outcome accurately measured to

7:44

minimize bias? So this is

7:46

where I'm a little bit unsure about.

7:48

The team came up with

7:50

their own definitions and

7:52

algorithm for assigning over-triage

7:54

and under-triage, and we'll talk

7:56

about this a little bit more in

7:59

our discussion. identified all important

8:01

confounding factors. I think

8:03

this is a little bit difficult because

8:05

there are so many factors that go

8:08

into the triage process and they were

8:10

reliant on variables that they

8:12

could easily pull from the

8:14

electronic health record. So there's

8:17

definitely a chance and they do address some

8:19

of this that they miss certain variables

8:21

or there were more subjective measures that

8:23

they were not able to capture. Do

8:25

you think the follow-ups of

8:28

subjects was complete enough? Yes.

8:30

How precise are the results?

8:32

Fairly precise. They had a

8:34

huge sample size, multi-center,

8:37

and had very tight

8:39

confidence intervals. And do you

8:41

believe the results? So, overall,

8:43

yes, but I think there

8:45

are some important caveats. You

8:47

know, it's not surprising that

8:50

E.S.I. triage is imperfect, and

8:52

there are problems with it.

8:54

But I am hesitant to

8:56

agree with the proportion of

8:58

their mistrage. Do you think the results

9:01

can be applied to the local

9:03

population? It depends on your local

9:05

health system and patient population.

9:07

Their study was focused primarily

9:09

in Northern California, and so

9:12

this may not be applicable to your

9:14

local practice setting if it's different.

9:16

Again, they were also based out

9:18

of the Kaiser Permanente health care

9:20

systems, so results may not be able

9:22

to be fully generalizable to

9:25

other hospital systems with different

9:27

triage protocols and cultures. And

9:29

lastly, the way they assign ESI

9:31

levels can differ from place to

9:34

place. For example, at my

9:36

institution right now, we routinely

9:38

will triage patients with sickle

9:40

cell disease or someone who

9:43

are concerned with fibromytropenia as

9:45

a level two, but per their

9:47

algorithm, they would triage them as

9:49

a level one. Yeah, and then also

9:51

I will add to this that the

9:53

ESI triage system is very common in

9:56

the United States, so... also depending on

9:58

if you practice in another. country,

10:00

you might be using a

10:02

different triage system as well. Brandon,

10:04

do you think the results of

10:06

this study fit with other available

10:09

evidence? Yes and no. It aligns

10:11

with the evidence that says that more

10:13

work needs to be done on ESI

10:15

and that ESI isn't perfect, but their

10:18

rate of mistriage around two-thirds

10:20

is higher than what was

10:22

previously noted. And finally, the

10:24

funding of this study, any

10:26

financial conflicts of interest here.

10:29

Many authors did receive

10:31

grants from Kaiser Permanente,

10:33

low-high-risk reduction program, and

10:35

Dr. Pines received funding

10:37

from some pharmaceutical companies

10:40

for unrelated work. Let's talk

10:42

about the results. So this

10:44

study included over 1 million

10:46

pediatric ED encounters across 21

10:48

emergency departments. The mean age

10:50

was 7.3 years, and there

10:52

were slightly more males than

10:54

females included. Brandon, what was

10:56

a key result. The key result

10:58

of the study is that the

11:01

ESI version 4 commonly mistriizes children

11:03

and that children were correctly triaged

11:05

only approximately a third at the

11:07

time, so a 34%. Breaking it

11:09

down by outcomes now, we'll have

11:12

a nice table for you regarding

11:14

the primary outcome, but as Brandon

11:16

mentioned... Correct tree out only happened

11:18

34% of the time. Most of

11:21

these kids were actually over triage

11:23

based on this study at 58.5%

11:25

and 7.4% were under triage. When

11:27

they looked at their secondary outcomes

11:30

and again those were the

11:32

patient and visit characteristics associated

11:34

with mistriage and those could

11:36

include age, sex, comorbidies, mode

11:38

of arrival and race and

11:41

ethnicity. So let's talk about

11:43

the patient and visit characteristics

11:45

that were associated with mistriage

11:47

because I actually found this

11:49

a little bit confusing. I think

11:52

I want to applaud the authors for

11:54

including this analysis for looking at

11:56

potential biases and stigma in triage

11:58

decisions because we know that this is

12:00

an important concept that's often

12:03

overlooked. And based on the results

12:05

of their study, it looks like

12:07

race and ethnicity, so Asian, black,

12:09

Hispanic, other, were more likely to

12:11

be both under and over triage

12:13

compared to white Caucasian. Under triage

12:15

was more likely if you were

12:17

older, so greater than six years

12:20

old, had other comorbidities or arrived

12:22

by ambulance or during normal office

12:24

hours, and over triage was more likely

12:26

if you were younger, so less than

12:28

three years. had no comorbidities or arrived

12:31

as a walk-in during off hours. There's

12:33

parts of this that make sense

12:35

to me and other parts that really

12:38

don't. I can see children of younger

12:40

age having the higher likelihood of getting

12:42

overtriaged, but I guess I don't quite

12:45

understand the comorbidity or arrival

12:47

parts of this because one

12:49

would think that if you had

12:51

more comorbidities, you would be at

12:53

higher risk of being overtriaged or

12:56

if you arrived by ambulance, maybe...

12:58

That would also bias us in

13:00

the assessment, right? And that patient

13:02

would be over triage as well.

13:05

But I think I'm just, I'm

13:07

getting too excited. I'm getting

13:09

ahead of myself here because

13:12

it is time for my

13:14

favorite section. Are you ready

13:16

to talk nerdy? Let's do it.

13:18

Let's get down to what first

13:20

nerdy point is. That triage

13:22

is messy. We have to acknowledge

13:25

that the process of triages,

13:27

by nature, imperfect. Traged

13:29

decisions are often made

13:31

very quickly, based on very

13:33

little information, with only a

13:35

cursory exam. So I don't

13:38

think it's realistic to be

13:40

accurate 100% of the time, and we

13:42

do have to give credit to our

13:44

nursing colleagues who do the triage,

13:46

because, man, it is hard. Yeah, it

13:49

is extraordinarily hard. constantly

13:51

impressed by their ability

13:53

to make these fast

13:55

quick and efficient decisions

13:57

with seemingly very little

13:59

information. They are potentially scanning

14:01

through the patient's chart and the

14:04

medical history to get a

14:06

very quick understanding of who they

14:08

are. This study tries to capture

14:11

that with a variable called

14:13

the Pediatric Comoridity Index, which is

14:15

based on the patient's prior ED

14:17

visits in patient stays ICU

14:19

utilization and coexisting illnesses and comorabilities.

14:23

We doubt that nurses in real

14:25

time are able to consistently access

14:27

this information as they're likely too

14:30

busy pulling these data points and

14:32

you know taking vital signs and doing

14:34

a physical exam. And patients or

14:36

families often report medical conditions

14:38

that may not always be

14:41

accurate in the triarch setting.

14:43

Oh yes, how many times have you seen?

14:45

Yeah, we got no comorbidities or no

14:47

other medical problems and you do a

14:50

chart review and you're like, whoa,

14:52

you have an underlying cardiac

14:54

condition? Huh, interesting. Or you ask

14:56

that what medication do they have? And they're

14:58

like, oh, I'm a lot of people, listen

15:00

to a pro. Yes, exactly. The list goes

15:03

on and on. So that's one

15:05

example of uncertainty and triage. Another

15:07

example of this uncertainty and triage

15:09

could be seen in their figure

15:11

under the level two section. They

15:13

ended up combining ESI two and

15:15

three because, quote, their algorithm did

15:18

not create rules for differentiating between

15:20

optimal assignments of all ESI two

15:22

and three visits. The other limitation

15:24

of their study in triage itself

15:26

is that the disease processes are

15:29

dynamic, right? They evolve over time

15:31

and in general modifying an

15:33

assigned ESI level is discouraged

15:36

But depending on how long

15:38

the patient has had to spend

15:40

in the waiting room Maybe their

15:42

initial assessment of ESI three or

15:45

four has gradually worsened and they

15:47

needed more resources and interventions The

15:49

authors did not report whether or

15:51

not they adjusted for this And if

15:54

the patient was initially triaged to a

15:56

lower ESI level and got sicker while

15:58

waiting, that doesn't necessarily mean... that the

16:00

initially assigned triage level was

16:02

inaccurate. Our second nerdy point

16:04

is about the variation in

16:06

ED pediatric volumes and nursing

16:08

triage experience or training. There

16:10

were 21 sites included in

16:12

this study with pediatric volumes

16:14

ranging from 3,000 to 25,000

16:16

visits per year and that's

16:18

quite an impressive range that

16:20

you can get in the

16:22

experience and training of the

16:24

nurses who performed this triage.

16:26

The authors did not specifically

16:28

detail the training of the

16:30

nurses, but they did state

16:32

that it ranged from a

16:35

minimum of four hours to

16:37

at least one year of

16:39

ED clinical experience. And, you

16:41

know, we would imagine that

16:43

both the volumes of pediatric

16:45

visits and the experience of

16:47

triage nurses contribute to inconsistencies

16:49

in triage decisions, but unfortunately

16:51

we can't see this subanalysis

16:53

based on experience or based

16:55

on specific practicing site. It's

16:57

going to be very unclear

16:59

whether or not these results

17:01

apply to our listener-specific setting.

17:03

And I will say Brandon,

17:05

I really don't think our

17:07

mistriage rate is as high

17:09

as 66%. Yeah, I agree.

17:11

Maybe my own personal mistriage

17:13

rate is as high as

17:15

66%. But our nurses do

17:17

a much better job. Thank

17:19

goodness for our nurses. Neri

17:21

point number three is about

17:23

their definitions for under and

17:25

over triage. This is the...

17:27

portion of the study that

17:29

I have a little bit

17:31

of concern with, one of

17:33

the reasons why the studies

17:35

may have different rates of

17:37

mistriage compared to previous studies

17:39

were their definitions for under-inover

17:41

triage. The authors developed their

17:43

definitions through a Delphi process

17:45

that included emergency medicine, pediatric

17:47

emergency medicine, and critical care

17:49

physicians, and emergency nurses. Their

17:51

system included critical interventions and

17:53

resource use. You can see

17:55

some examples of this in

17:57

their supplement. But we do

17:59

want to give them credit

18:01

because they did several rounds

18:03

of manual record review to

18:05

validate these definitions. But this

18:07

was a group of eight

18:09

people. Would other physicians, researchers,

18:11

agree with these definitions? Are

18:13

there some of their metrics

18:15

that are more institution or

18:17

health care system specific? And

18:19

you know, this process is

18:21

difficult because what do you

18:23

determine as the best outcome

18:25

to determine appropriate triage? And

18:27

this is this idea, this

18:29

concept is still heavily debated.

18:31

And so everyone is going

18:33

to come to a different

18:35

consensus and conclusion for this.

18:37

Well, I think nerdy point

18:39

number four adds a little

18:41

bit more to this confusion

18:43

in the complexity, right? And

18:45

that's around their consultations and

18:47

procedures. The authors, they didn't

18:49

include any specialty consultations. or

18:51

simple procedures, such as laceration

18:53

repairs or fully catheters, as

18:55

resources used, as they were

18:57

not consistently available as data

18:59

points during the course of

19:01

their study. It's unclear how

19:03

this may have affected their

19:05

overall over and under triage

19:07

rates, because these additional resources

19:09

were not included, it's possible

19:11

that some of the kids

19:13

were deemed over triaged, were

19:15

actually appropriately triaged, because they

19:17

needed these. repairs, these consultations,

19:19

and it's also possible that

19:21

some children who are deemed

19:23

appropriately triage were truly under

19:26

triage. And finally, nerdy point

19:28

number five is about clinically

19:30

significant mistriage. The study focuses

19:32

on triage accuracy rather than

19:34

patient-oriented outcomes such as mortality

19:36

or ICU admission. And not

19:38

all mistriage is equivalent. So,

19:40

say for example, we see

19:42

that in their study, there

19:44

were a portion of ESI

19:46

level 4 patients who are

19:48

classified as ESI 5. But

19:50

in a real clinical setting,

19:52

does an ESI 4 versus

19:54

a 5 really matter? Does

19:56

harm come to a patient

19:58

for being labeled as an

20:00

ESI 5? when they were

20:02

four, that they require one

20:04

more resource that anticipated, is

20:06

that really such a bad

20:08

thing? And a lot of

20:10

institutions, ESI level four and

20:12

fives, are being co-hearted into

20:14

a fast-track or an urgent

20:16

care setting, and did it

20:18

really change the overall course

20:20

of their management? Now in

20:22

contrast, that same ESI level

20:24

four or five patient that

20:26

you mentioned, if they ended

20:28

up being a level one

20:30

or two who needed more

20:32

timely interventions, even admission. Now

20:34

those are examples of clinically

20:36

significant mistrias that can actually

20:38

impact our patient-oriented outcomes or

20:40

who's which we love. Exactly,

20:42

exactly. So Brandon, can you

20:44

comment on the author's conclusion

20:46

compared to the SGM conclusion?

20:48

We agree with the author's

20:50

conclusion. Now while we believe

20:52

that ESI levels are not

20:54

always accurate for treating children,

20:56

We're not sure about the

20:58

true rate of mistriars that

21:00

is reported in this study.

21:02

And give us the S

21:04

gem bottom line. We need

21:06

better and more accurate methods

21:08

for triaging children in the

21:10

emergency department. That case that

21:12

you presented at the beginning,

21:14

the medical director that's coming

21:16

to you and asking you

21:18

about how do we triage

21:20

pediatric patients, what are you

21:22

telling them? I'm telling them

21:24

that hey, ESI levels are

21:26

not always accurate, especially for

21:28

pediatric patients. Mistriage has clinically

21:30

significant consequences. Rates of mistriage

21:32

is highly variable based on

21:34

triage nurse experience and training.

21:36

And at this time, I

21:38

recommend directing efforts towards identifying

21:40

factors in previous instances when

21:42

children were mistriage that resulted

21:44

in harm and focusing efforts

21:46

to address these factors. So

21:48

Brandon, how are you applying

21:50

the findings of this study

21:52

clinically? ESI levels are the

21:54

most popular method of ED

21:56

triage in the US. At

21:58

this point, we're not getting

22:00

away from ESI, but we should

22:02

recognize that they're not always accurate

22:05

and that there's a lot of

22:07

potential biases and inaccuracies inherent

22:09

in the process that could

22:11

lead to mistriage rates. And

22:13

it's important that we continue

22:15

to strive towards accurate triage

22:18

because it helps us to

22:20

prioritize patients, allocate resources efficiently

22:23

and effectively. And when we

22:25

evaluate our triage accuracy, we should

22:27

focus not on. just mistriage rates,

22:30

but also on what is considered

22:32

clinically significant mistriage. Brandon, you

22:34

know a lot about triage and you've done

22:36

a lot of reading and looking into this.

22:38

Do you want to tell us a bit

22:40

about the work that you're doing? Yeah,

22:43

absolutely. So even in this study, we've

22:45

identified some problems with our current triage

22:47

system, and a lot of it boils

22:49

down to the fact that we are

22:52

making very quick decisions on very

22:54

little information. And so how

22:56

can we expand the real-time

22:58

information to make more accurate

23:01

triage decisions? And so my research

23:03

focus right now has been looking

23:05

at leveraging AI in

23:08

pediatric emergency medicine. And

23:10

my hope is that we

23:12

can leverage large language models,

23:14

generative AI, machine learning, to

23:16

help summarize information and

23:19

predict ESI levels. We've

23:21

published a comparison. of

23:24

the accuracy and reliability of

23:26

various foundational models, such as

23:29

CHATGBT, Claude, at predicting

23:31

ESI levels, and we've shown

23:33

that with clinical vignettes, we can

23:35

achieve an accuracy as high as

23:37

74 to 82 percent. Now, we

23:39

understand that these are just

23:42

clinical vignettes, so we are

23:44

working on a multi-center retrospective

23:46

study, looking at patient data

23:48

to evaluate the accuracy, the

23:50

reliability of AI triage. And

23:53

our hope is not that our AI

23:55

overlords are going to come in and

23:57

replace Nurs triage, but that a concern...

24:00

as a clinical decision support tool

24:02

to improve overall accuracy and efficiency.

24:04

And I really like this term,

24:06

augmented intelligence, where we are using

24:08

AI to support nurses in real

24:11

time to do things that they

24:13

normally could it, namely synthesizing and

24:15

analyzing large amounts of a patient's

24:17

history, presentation, vital signs, to predict

24:20

a more accurate ESI level. Brandon,

24:22

I have to say, I look

24:24

forward to the day that this

24:26

is not just available for triage,

24:29

but also for us as clinicians,

24:31

right, in the emergency department to

24:33

be able to tell an AI

24:35

to help us, hey, give me

24:37

a nice summary of the things

24:40

that I cannot miss about this

24:42

patient in their medical history. Man,

24:44

that would be awesome in comparison

24:46

to just clicking through multiple clinic

24:49

notes, previous ED notes, hospital discharges.

24:51

Oh, man, I'm looking forward to

24:53

that day. I would sell my

24:55

left leg for that. All right

24:58

Brandon, let's finish it off here.

25:00

What are you telling your medical

25:02

director? So medical director. The ESI

25:04

system is one of the most

25:06

popular in the United States and

25:09

you are right that it isn't

25:11

always accurate for the triage of

25:13

pediatric patients. I don't know of

25:15

any specific triage system to recommend

25:18

over the ESI system currently as

25:20

the triage process itself is challenging.

25:22

because we are making decisions based

25:24

on such little information in a

25:27

very short time period. I think

25:29

we should look at the cases

25:31

of mistriage that result in clinically

25:33

significant harm or clinically significant outcomes

25:35

and see what factors may have

25:38

impacted those. In the meantime, I'm

25:40

going to keep an eye out

25:42

for any further studies that may

25:44

suggest that there's a better way

25:47

to triage kids. Hopefully in the

25:49

future we have some additional AI

25:51

or guidance to help us with

25:53

our triage accuracy. Well Brandon, thank

25:56

you so much for joining us

25:58

on S gem beads. Thank

26:00

you so much for having me. It's

26:02

been a great time, Dennis. a great time,

26:04

And before we go, we go, can you

26:06

give us the SGM tagline? All right, listeners, remember to remember

26:08

to be skeptical of anything you even if

26:11

even if you've heard it on the Skeptics

26:13

Emergency Medicine. Talk to

26:15

everyone next time. time.