8/8/16 Clinical Decision Making - Willett

Question 1

steps involved in clinical reasoning

Answer 1

1. H&P : pattern recog, come up with ddx, think of pretest probabilities
2. diagnostic testing : calc post-test prob
3. decision re: tx (threshold concept)
4. re-evaluation

Question 2

threshold concept

Answer 2

“how sure is sure enough?”

multiple variables to consider

• outcome of treated vs untreated disease
• conseq of false pos/neg tests
• benefit/risk/cost of dx tests, tx,
• outcome of urgent vs delayed tx

Question 3

positive vs negative thresholds

Answer 3

negative threshold

how sure that someone DOES NOT have disease to be comfortable NOT treating

positive threshold

how sure that someone DOES have disease to be comfortable treating

Question 4

factors affecting negative threshold

factors affecting positive threshold

Answer 4

NEGATIVE

• untreated disease having v serious/fatal outcome
• tx being very effective
• early tx far better than late tx

ex. meningitis!

POSITIVE

• tx is toxic, risky
• prognosis is poor
• definitive dx test is risky

Question 5

therapeutic decision making and PICO

Answer 5

evaluate treatments by looking at studies through PICO

Patient

Intervention

Comparison (what’s the intervention being compared to?)

Outcome

Question 6

PICO pitfalls

Answer 6

Patients: sicker (referral) or less sick (volunteers) than avg

Intervention: int might be specialized/unavailable for your pt

Comparison: look out for comparisons that are set up to fail and/or for inadequate placebo/blinding

Outcome: look out for composite/surrogate outcomes (vs. ones that are relevant/imp for your pt)

• composite - lots of outcomes bound up in one outcome
• surrogate - something we measure vs something that actually matters (an analyte level vs result that matters to pt)

Question 7

how great is the benefit???

how great is the risk???

how sure am I of these numbers?

what should you be looking at for each of these?

Answer 7

how great is the benefit???

RRR vs. ARR, NNT

how great is the risk???

RRI vs. ARI, NNH

how sure am I of these numbers?

confidence interval

Question 8

ex. tx reduces bad outcome from 4% to 3%

ARR

RRR

NNT

Answer 8

ARR = 1%

RRR = 25%

NNT = number of people treated to prevent 1 death = 1/ARR = 100

Question 9

ex. tx reduces bad outcome from 10% to 15%

ARI

RRI

NNH

Answer 9

ARR = 5%

RRR = 50%

NNH = 1/ARI = 20

Question 10

criteria for a reasonable screening test

Answer 10

• substantial burden of suffering from target disease
• test is a fairly good predictor of presence/absence of disease
• finding and treating early gives better outcomes
• benefits of screenings outweigh harms of screening!

Question 11

LR+

LR-

Answer 11

likelihood of + test in diseased / likelihood of + test in NON-diseased

sn / 1 - sp

(want it to be >10)

likelihood of - test in diseased / likelihood of - test in NON-diseased

1 - sn / sp

(want it to be <.1)

Question 12

over-diagnosis

Answer 12

diagnosis of an issue (that then recommends tx) that would never affect pt in their lifespan