Evidence-Based Pediatrics for the Real World

advertisement
Evidence-Based Pediatrics
for the Real World:
Reading, Reflecting, and ‘Rithmetic
Bud Wiedermann, MD, MA
Professor of Pediatrics, George
Washington University
Attending in Infectious Diseases,
Children’s National Medical Center
Washington, DC
Faculty Disclosure
My spouse and I have not had any
relevant financial relationships
during the past 12 months.
Educational Need/Practice
Gap
• The Problem
– EBM = “empiricist quackery”?
– Barriers of busy practice, volume of
new information
• The Solution
– Demystify EBM
– Relax, you don’t need to read
everything!
When you walk out of here,
you will be able to….
• Restate clinical conundrums as
answerable questions
• Effectively and efficiently find
answers to clinical conundrums
• Relate p values, likelihood ratios,
and number needed to treat to
real-world, practical pediatric
practice.
Reading
‘Rithmetic
Reflecting
Reading Has 2 Components
Reflection Has 3
Components
• In action
• On action
• For action
‘Rithmetic Has Too Many
Components!
• Frequentists
vs. Bayesians
• Number
needed to
treat
• Likelihood
Ratios
Time for a Case
Caitlyn is a 22 month-old girl,
previously healthy except for minor,
self-limited viral illnesses who
presents with a 36 hour history of
fussiness, occasional tugging at her
left ear, and fever to touch. Exam
reveals a slightly fussy infant with a
temperature of 38.7 C, left middle
ear effusion, and a slightly bulging,
erythematous TM.
Do You Have a Question?
• Background
– Any recent
textbook
– MD Consult
– Up-to-Date
– Etc.
• Foreground
– PubMed
– Up-to-Date?
– Dynamed?
– Google?
“If they can get you asking the
wrong questions, they don't have
to worry about answers.”
-Thomas Pynchon
Take a Shortcut!
PICO
Another Caveat About
Reading the Literature
• Don’t forget to browse!
– How do you know what you need to
know?
– Curiosity kills cats, not physicians!
• Primary sources
– NEJM, JAMA, BMJ
– Pediatrics
• Pediatrics
in Review
• AAP Grand
Rounds
• EvidenceBased
Child
Health
If You Find a Good One, Ask
Yourself ….
• Could Caitlyn have been included
in the study?
• If not, is she close enough that
can I still use some of the
results?
– If yes, read the Methods and Results
sections (not the abstract or
conclusions)
– If no, put it aside for another time (or
not)
Rx of AOM In Children < 2 yo
• Inclusion criteria
– AOM Severity of Symptoms Score >3
(Pediatr Infect Dis J 2009;28:9-12)
– Presence of middle ear effusion
– Moderate or marked bulging of TM,
or slight bulging with either otalgia
or marked erythema of membrane
• RCT – high dose amox/clav vs.
placebo
NEJM 2011; 364:105-15
Clinical Failure Higher in
Placebo Group (‘Rithmetic)
Visit
Day
4-5
10-12
Antibiotic Placebo
p
NNT
4%
23%
<0.001
5
16%
51%
<0.001
3
NNT: What is it Good For?
(Absolutely Nothing?)
• I, like most physicians, don’t do a
good job of explaining risk to
patients and families
• The second part of Reading
– Read Your Patients
– Understand Their Values
• NNT = 5 with the right side of your
brain
J Gen Intern Med 2008; 23:2117-2124
Health Psychol 2009;28:21-216
BMJ 2003; 327:745-748.
Risk Characterization
Theater
Rifkin and Bower (2007). The Illusion of
Certainty. New York, Springer.
If You Can’t Draw….
…. Foster Insight
Which of these works better for
communicating single event
probabilities?
You have a 30% chance of a side
effect from this drug.
Three out of every 10 patients
have a side effect from this drug.
BMJ 2003;327:741-744.
4/1000 women >40 yo who do not
undergo breast cancer screening
die of breast cancer, vs. 3/1000 of
those who do.
In every 1000 women who undergo
screening one will be saved from
dying of breast cancer.
Mammography reduces breast cancer
by 25%.
To prevent one death from breast
cancer, 1000 women need to undergo
screening for 10 years.
Accuracy of Pneumatic
Otoscopy (More ‘Rithmetic)
• Sensitivity 87%
• Specificity 89%
• Is that good enough?
– If yes, what is it good for?
– If no, why not?
• What do sensitivity and specificity
mean to you?
Arch Dis Child 1992; 146:433-435
Playing the Horses (Odds)
• LR+ = sens/(1-spec)
= 0.87/0.11 = 7.9
• LR- = (1-sens)/spec
=0.11/0.89 = 0.12
Reflecting
• Our individual experiences are a
form of evidence.
• Subject to errors
– Pretest probability
– Strength of evidence
BMJ 2002;324:729-732.
Availability
• We overestimate
frequency of vivid
or easily recalled
events.
• This is why we
love zebras!
Representativeness
• Judging how
similar a case is
to a prototype
• Usually
overestimates
probability
• … and its pal,
Anchoring
…. And more
• Wrong decisions based on:
– How detailed the description is
(Support Theory)
– Order that the data are
presented (later data=more
weight)
– Tendency to overestimate
probability of serious but
treatable diseases
Other Resources
Otitis Media
• NEJM 2011;364:116-126.
• Management of Acute Otitis
Media: Update. Evidence
Report/Technology Assessment
No. 198. RAND/AHRQ, November
2010.
http://www.ahrq.gov/clinic/tp/otitis
uptp.htm
(Warning: 207 pages)
Other Resources
EBM
• University of Alberta
– http://www.ebm.med.ualberta.ca/
• University of Oxford
– http://www.cebm.net/
• JAMA Users’ Guides
– http://pubs.amaassn.org/misc/usersguides.dtl
Questions?
bwiederm@cnmc.or
g
Download