Risk Or, Hippocrates was wrong P. Mukherji. Primum non nocerum -HIPPOCRATES.

RiskOr, Hippocrates was wrong

P. Mukherji

Primum non nocerum-HIPPOCRATES

First do no harm FIRST: implies that this is a

cardinal and overarching tenet of medicine

The statement also implies that this is possible.

Better motto:

Please try and kill as few patients as possible, while hopefully healing and helping as many as possible.

Please stop me Please interrupt Please look it up

91 yo WM s/p trip and fall +head injury: abrasion NT head/spine, full ROM No concerning sx, no

ASA/Coumadin

CT?

RULES to help us? Canadian Head CT

NEXUS II

New Orleans Criteria

To scan or not to scan?

Spin

em

all

and le

t Go.

..

Nope,

he

mig

ht get

a ..

.

0%0%

a. Spin em all and let God sort em out

b. Nope, he might get a brain tumor

CT shows small SAH

SDH is

mor

e co

...

Epid

ural i

s m

o...

He

has a

n aneu

...

He

needs

ICU t.

..

He

needs

a m

on...

0% 0% 0%0%0%

a. SDH is more common in trauma

b. Epidural is more common in trauma

c. He has an aneurysm

d. He needs ICU timee. He needs a monitor

Pt. admitted for observation Falls off the bed(!) When transferred back to

stretcher stops breathing Regains VS, is transported to CT Stops breathing again!

CT shows? C-spine is crumply at C2

CT head is clean Attending read later finds initial CT

finding to be motion artifact

Did we do wrong? Pt. was admitted for his own safety Harm came to patient Admission led to harm

Should we change practice? CO- vs. O-MISSION

Would you do it differently next time?

Hea

d CT

plea

se

How a

bout

jus.

..

0%0%

a. Head CT pleaseb. How about just

good instructions

Doctors are biased*

Routinely overestimate benefits of intervention

Routinely minimize risks of intervention

*Surgeons!

House of God? FAT MAN’s RULE #13

“The delivery of good medical care is to do as much nothing as possible”

But it’s not just about doctors We all routinely underestimate

everday risks High frequency, unlikely events

We tend to demonize and overestimate rare events Low frequency, very unlikely events

Did you ever fall? Over your lifetime, falling is a

significant risk

Falls in the elderly are a problem, should you be attentive to it already?

What’s most likely to kill ya?

Shar

k at

tack

MRSA

Flu

Fal

l

25% 25%25%25%a. Shark attackb. MRSAc. Flud. Fall

Actual lifetime odds of dying Sharks: 1 in 60,453

Falls: 1 in 218

MRSA: 1 in 197

Flu: 1 in 63!

Risk vs. Benefit We accept the risk of anaphylaxis

when prescribing antibiotics,

AS LONG AS

We think there is some benefit to the antibiotics for the pt.

Overutilization? Overuse is defined by multiple

federal overseers and review boards as: Testing for which no (or minimal)

benefit to the patient exists

Who has prescribed Abx?

…for sinus pressure? …for cough/bronchitis? …for sore throat?

31M with fever and sore throat, has exudates, tender ant. nodes, and no cough a. Bicillin shot!b. Z-packc. Rapid strep, Cx

if negd. Rapid strep, do

what is sayse. Suck it up, wuss

It’s cool, you haven’t killed (Probably)

But you haven’t helped

…and you might have killed

Sore throats 14 million visits in U.S./yr. Steroids/NSAIDS >> Abx NO evidence that it helps abscess ONLY evidence on rheumatic heart

from worst military outbreak ever

NNT? 40,000 NNH? 5 minor rxn, 6 to recur, 2500

majorSmartem.org, David Newman, AEM 2010

Let’s talk about Our cognitive errors

Assessing and communicating risk

Balancing risk and intervention

Unintended Consequences**

Interventions ALWAYS have the potential to create unanticipated and unforeseen events Perverse

Every intervention/test carries RISK and UNCERTAINTY RULE #1

LAW of UNINTENDED CONSEQUENCES

We are expected to detail these risks to our patients Informed consent: risk/benefit-

uncertainty on both ends

Risk assessment

We suck at this

Probability of Occurrence x Impact of Risk Event

We suck at probability, too. Statistics, anyone?

You test a bunch of people for HIV

1 is positive Likelihood of true positive?

Math! Test is 99.99% sensitive AND

specific

Out of 10,001 men, 1 has HIV

So what do you do with your positive guy?

Sir, your rapid test was positive, you need a repeat test but…

It’s

ver

y lik

e...

It’s

ver

y unli.

..

Pro

bable

, but .

..

Pro

bably

wro

ng

25% 25%25%25%a. It’s very likely correct

b. It’s very unlikely to be correct

c. Probable, but could be wrong

d. Probably wrong

Great for negatives (screen)

One pt. with HIV will be positive.

One other pt. will have a positive test.

50% chance that this is a true positive** PPV = 50% (RULE #2)

Testing?

On low prevalence groups leads to higher rates of false positives**

RULE #2

Screening Hgb A1C

Mammography

PSA

PSA

Screening PSA will result in an absolute mortality reduction of 33%

PSA 17 of 100 men will get a dx of CA

3 will die if untreated Treatment will save the life of 1 of

3* Treatment will kill 1 of the 17 10 of the 17 will be incontinent

and/or impotent

How you present the data matters

“98.5% safety from a particular disease”

That is exactly 1 in 63, the odds that you’ll be killed by flu in your lifetime.

How likely are you to pass the inservice?

50%

90%

95%

98%

99%

20% 20% 20%20%20%a. 50%b. 90%c. 95%d. 98%e. 99%

You did not grow a brain that likes small or huge numbers

All your brain sees is either a really high (98%, 99%) likelihood and rounds up

or a really low (1%, 2%) and rounds down

Probabilities are percentages

But real numbers work MUCH better RULE #3: USE REAL NUMBERS Want someone to go home?

“98% you’re ok!” Want someone admitted?

“1 out of 50 you’re dead!”

How do some docs manage risk?

TESTS We’re even told that pts. WANT tests.

Testing does not reduce legal action and might increase it

Every test is an additional intervention** RULE #1 Failure to follow up Failure to interpret correctly Failure to pursue to the correct test Alteration of the presenting frame Incidentaloma Radiation

Prevalence of incidental findings in trauma patients detected by computed tomography imaging

>3000 pts. 990 (32%) had Type I findings 1274 (41%) had Type II findings

631 incidentalomas concerning for nodules, masses

Incidence and Predictors of Repeated CT-PA in ED patients

Longitudinal study of 675 pts. 33% had repeat CT-PA 5% had >5 repeat CT-PA 75% had some CT scanning

Limitations: possible under-reporting

Don’t just stand there… Inherent predisposition to

intervening

Lack of tolerance for uncertainty

Errors of comission and omission are treated differently

Testing does not transfer risk

Most “high utilizers” are not driven by litigation fears

Most cite diagnostic concerns, “best for the pt.”, “don’t want to miss anything”

“Not missing anything” Implies need for diagnostic

certainty Relays fears of sudden unexpected

morbidity

Outcome based practice Low risk patients may not get a dx

Who’s the most conservative Jacobi attdg?

Per

rera

Gru

ber

Corc

iari

Hau

ghey

Jones

Sufa

i

17% 17% 17%17%17%17%a. Perrerab. Gruberc. Corciarid. Haugheye. Jonesf. Sufai

Most clinically cowboyish?

Per

rera

Gru

ber

Corc

iari

Hau

ghey

Jones

Sufa

i

17% 17% 17%17%17%17%a. Perrerab. Gruberc. Corciarid. Haugheye. Jonesf. Sufai

What if all the cowboyish guys worked fasttrack and main?

And the most conservative attdgs worked resus/trauma?

Poker, anyone?

Know thyself

Questions? We a low level of comfort with

uncertainty; it makes us want to do stuff

Testing low risk groups has problems

Discussion of risk and benefit (to whom and of what) is ongoing. Minimalists will win this discussion.

Take home We should expect much clearer

thinking about risk to our pts.

We should accept that we are responsible for both our actions and inactions, and tailor approach to pt.

Thank you

Dr. Schriger, UCLADr. Gallagher, MontefioreDr. Newman, Sinai

choosingwisely.orgTheNNT.combestbets.orgCochrane Review

@ercowboy