This document discusses translating evidence into practice in the era of electronic medical records. It provides an overview of the translational pathway from biomedical research to improved population health. On average, it takes 16 years to translate a biomedical innovation from the bench to the bedside. Several factors influence the speed of adoption of new practices, including consciousness-raising, changing social norms, and making new behaviors easier through system changes. The document reviews sources of evidence for implementation, including local experts, journals, Cochrane reviews, professional societies, and clinical practice guidelines. Trustworthy guidelines are developed through a transparent process and are based on systematic reviews.
Crotty engaging patients in new ways from open notes to social media
Cost effectiveness & evidence-based medicine
1. Transla'ng
(cost-‐effec've)
evidence
into
prac'ce
in
the
EMR
era
Andrew
D.
Auerbach
MD
MPH
Professor
of
Medicine
UCSF
Department
of
Medicine
Chair,
UCSF
Apex
Clinical
Content
Oversight
CommiKee
2. Overview
• Transla'on
and
implementa'on
of
evidence
• Where
to
get
evidence
for
those
who
want
(need)
to
implement
– Transla'onal
issues
in
EMR’s
– Evidence
Acquisi'on,
implementa'on
• Costs,
value,
and
cost-‐effec'veness
3. The
Transla'onal
pathwayascade
Improved
Bench/ Bench to Clinical Comparative Comparative
Implement population health,
Biomedical bedside efficacy effectiveness effectiveness
practices Improved
Research translation knowledge research knowledge
effectively healthcare value
Determine how patient, provider, and delivery
Determine causal Determine associations
system changes influence outcomes
pathways between treatments and
outcomes - Health system redesign
Outcomes and health services - Scaling and dissemination of delivery system changes
Clinical efficacy trials
research
- Research in redesign and dissemination
4. It
takes
a
long
'me
to
translate…
• On
average,
it
takes
16
years
to
move
a
biomedical
innova'on
from
bench
to
bedside
• It
is
unknown
how
long
it
takes
to
have
a
prac'ce
used
effec'vely
5. Why
so
long?
• Unaware
of
a
Cons
of
changing
outweigh
the
pros
at
the
top
Precontempla'on
problem,
not
intending
to
make
a
of
this
cascade
change
• Recogni'on
of
Contempla'on
problem,
beginning
to
look
at
pros
and
cons
of
change
• People
ready
to
take
Prepara'on
ac'on,
may
require
assistance
• People
have
Ac'on
modified
behaviors
to
address
problem
• People
able
to
sustain
ac'on,
The
pros
surpass
the
cons
for
later
stages
Maintenance
working
to
prevent
relapse
..But
backsliding
can
always
take
place
6. Factors
which
speed
adop'on
• Consciousness-‐Raising
– Increasing
awareness
via
informa'on,
educa'on,
and
personal
feedback.
• Changing
social
norms:
– What
makes
the
new
behavior
cool
and
the
old
one
uncool?
– Is
the
preferred
behavior
an
important
part
of
who
we
are
and
want
to
be?
– Does
the
old
behavior
affect
others
nega'vely?
– Are
people
generally
suppor've
of
change?
• Counter-‐Condi'oning
– Subs'tu'ng
healthy
ways
of
ac'ng
and
thinking
for
unhealthy
ways
– Provide
rewards
for
the
new
behavior,
and
eliminate
rewards
related
to
old
ways
• Self-‐Efficacy
– Believing
in
one’s
ability
to
change
and
making
commitments
to
act
on
that
belief
• System
change:
– Make
the
new
behavior
easier
to
carry
out
than
the
old
one
– Provide
reminders
and
cues
that
encourage
the
new
behavior
7. Where
to
get
evidence
• Local
experts
– Advantages:
Local
champion,
may
represent
local
priori'es,
may
have
a
‘how
to’
component
– Disadvantages:
May
not
be
truly
evidence-‐based
o
• Latest
NEJM
(or
pick
your
journal)
– Advantages:
High
face
validity
– Disadvantage:
no
‘how
to’,
always
another
study
on
the
way
8. Where
to
get
evidence
• Cochrane
database
of
systema'c
reviews
– hKp://www.cochrane.org/cochrane-‐reviews
– Advantage:
Comprehensive
synthesis
of
evidence,
explana'on
of
where
evidence
is
clear/unclear
– Disadvantage:
No
clear
recommenda'ons
for
how
to
use
the
evidence
• Professional
socie'es
– Advantages:
Clear
recommenda'ons
– Disadvantages:
Poten'al
COI,
varia'on
across
compe'ng
socie'es
9. Clinical
prac'ce
guidelines
• A
specific
subset
of
evidence
– Considered
a
‘standard’
source
of
prac'ce
evidence
– Recent
controversies
(mammography,
PSA,
CT
Angio,
etc)
prompt
discussion
regarding
what
comprises
a
‘trustworthy’
clinical
prac'ce
guideline
(CPG)
10. Which
guidelines
should
you
choose?
•
To
be
trustworthy,
guidelines
should:
– Be
based
on
an
explicit
and
transparent
process
that
minimizes
distor'ons,
biases,
and
conflicts
of
interest;
– Be
based
on
a
systema'c
review
of
the
exis'ng
evidence;
– Be
developed
by
a
knowledgeable,
mul'disciplinary
panel
of
experts
and
representa'ves
from
key
affected
groups;
11. Which
guidelines
should
you
choose?
•
To
be
trustworthy,
guidelines
should:
– Consider
important
pa'ent
subgroups
and
pa'ent
preferences,
as
appropriate;
– Provide
a
clear
explana'on
of
the
logical
rela'onships
between
alterna've
care
op'ons
and
health
outcomes,
and
provide
ra'ngs
of
both
the
quality
of
evidence
and
the
strength
of
recommenda'ons;
and
– Be
reconsidered
and
revised
as
appropriate
when
important
new
evidence
warrants
modifica'ons
of
recommenda'ons.
12. What
should
recommenda'ons
include
• For
each
recommenda'on,
the
following
should
be
provided:
•
Explana'on
of
reasoning
underlying
the
recommenda'on,
including:
• Descrip'on
of
poten'al
benefits
and
harms.
• Summary
of
available
evidence
(and
gaps),
descrip'on
of
the
quality
(including
applicability),
quan'ty
(including
completeness),
and
consistency
of
evidence.
• An
explana'on
of
the
part
played
by
values,
opinion,
theory,
and
clinical
experience
in
deriving
the
recommenda'on.
13. What
should
the
recommenda'ons
include
• For
each
recommenda'on,
the
following
should
be
provided:
• A
ra'ng
of
the
level
of
confidence
in
(certainty
regarding)
the
evidence
underpinning
the
recommenda'on.
• A
ra'ng
of
the
strength
of
the
recommenda'on
.
• A
descrip'on
and
explana'on
of
any
differences
of
opinion
regarding
the
recommenda'on.
14. What
the
IOM
did
not
talk
about
• How
to
translate
guidelines
into
EMR
decision
support
or
ordersets
–
LiKle
informa'on
on
how
to
implement
CPG’s
in
a
way
that
is
concordant
with
core
recommenda'ons
•
Most
likely
affects
complex
algorithmic
decision
support,
such
as
early
warning
systems
– ‘Atomiza'on’
of
evidence/decision
making
in
EMR’s
• The
workflow
in
EMR’s
is
usually
fundamentally
different
than
that
envisioned
by
a
CPG.
15. EMR’s
provide
the
needed
connec'ons
All
these
connec'ons
can
be
made
by
the
EMR,
but
are
selected
by
people
16. Humbling
evidence
• EMR’s
for
improving
health
quality:
– Small
to
moderate
sized
improvement
in
acute
care
process
measures,
no
impact
on
outcomes
in
36
published
studies*
(Sahota,
Implementa'on
Science
2011)
– Heterogeneous
impact
on
management/screening
of
chronic
condi'ons
(Roshanov,
Implementa'on
Science
2011)
*Don’t
feel
badly,
most
of
QI
is
in
the
same
boat
17. Costs
Value–
the
next
fron'er
• To
increase
value
you
must
tackle
costs
– “Approval
driven”
approaches
• P&T
CommiKees,
an'microbial
stewardship
– System
redesign
• Six
Sigma/Lean
Sigma
– Some
ques'ons
for
the
future
19. 50,000
Some
diseases
have
goKen
more
costly
faster
45,000
40,000
35,000
MI
Hospital Charges ($)
30,000
CHF
CAP
25,000 COPD
UTI
CVA
20,000 Sepsis
15,000
10,000
5,000
0
1993 1995 1997 1999 2001 2003
Year
Rothberg
M,
Health
Aff
(Millwood).
2010
Aug;29(8):1523-‐31.
20. 20%
What
are
we
gerng
for
our
money?
18%
16%
14%
In Hospital Mortality
12% MI
CHF
CAP
10% COPD
UTI
CVA
8% Sepsis
6%
4%
2%
0%
1993 1995 1997 1999 2001 2003
Year
Rothberg
M,
Health
Aff
(Millwood).
2010
Aug;29(8):1523-‐31.
21. Where
are
the
cost
reduc'on
opportuni'es?
Missed
Prev
Opps,
Prices
That
Are
Fraud,
$55
Too
High,
$105
$75
Inefficiently
Delivered
Services,
$130
Excessive
Administra've
Costs,
$190
Unnecessary
Services,
$210
Non-‐Wasteful
Spending,
$1,735
Low
hanging
fruit
of
inefficient
and
wasteful
care
are
present,
but
larger
benefit
may
be
elsewhere
22. Cost
types
in
healthcare
• Fixed
costs
– Costs
that
do
not
vary
over
ranges
of
output.
• Buildings/Equipment
–
Paid
for
once
• Salaried
personnel
–
paying
for
anyway
– For
example:
• A
PET
scanner
is
expensive,
but
it
is
paid
for
once
• The
cost
of
upkeep,
space,
and
the
PET
technician
don’t
vary
substan'ally
as
more
people
use
it
23. Point
of
clarifica'on
• Where
do
guidelines/pathways
fit
in
this
talk?
– Lots
of
studies
on
guidelines,
pathways
– Minority
report
costs
as
an
outcome
– Few
used
guidelines/pathways
with
the
aim
of
reducing
costs/u'liza'on
(and
few
succeeded).
24. Cost
types
in
healthcare
• Variable
costs
– Costs
that
change
as
the
volume
of
services
increases
• Some
medica'ons,
material
costs
– For
example:
• Reducing
the
number
of
CT
scans
may
reduce
the
amount
of
contrast
purchased
and
used.
25. Cost
types
in
healthcare
• Marginal
costs
– Elsewhere
in
the
world:
• Costs
to
produce
an
addi'onal
product
decrease
with
each
addi'onal
unit
– In
healthcare:
Not
usually
the
case
• Cost
per
unit
output
fixed
(See
Fixed
costs)
un'l
maximal
capacity
reached
• Addi'onal
PET
scanner
or
PET
scan
not
priced
lower
than
the
first
one.
• Replacement
op'ons
generally
not
of
lower
cost
(think
Xa
inhibitor
vs.
warfarin)
26. Why
is
Econ
101
in
an
EMR
talk?
• Vast
majority,
70-‐80%
-‐
maybe
as
high
as
84%
of
costs
in
health
care
are
fixed
costs
– Building
upkeep,
equipment,
personnel
occupy
bulk
of
costs
– Variable
salaries
and
discre'onary
items
(e.g.
drugs,
materials)
represent
a
small
propor'on
– EMR’s
• Implica'ons
over
the
short
term:
– Reducing
variably
costed
items
will
have
limited
impact
– Efforts
to
reduce
u'liza'on
of
costly
items
oxen
offset
by
compensatory
efforts
to
maintain
revenue
to
subsidize
fixed
costs
– Goal
will
need
to
focus
on
reducing
fixed
and
variable
costs
in
tandem
Roberts
R
JAMA,
February
17,
1999—Vol
281,
No.
7
27. Why
is
Econ
101
in
an
EMR
talk?
• Implica'ons
over
the
longer
term:
– EMR’s
used
to
automate
human
tasks
– EMR’s
used
to
eliminate
need
for
fixed
cost
items
– Must
provide
clear
cost
and
u'liza'on
data
– Overcome
barriers
or
innovate
on
old
models
• Physician
awareness
• Health
technology
acquisi'on
commiKees
• P&T
• An'microbial
stewardship
cmte
• Pathways
Roberts
R
JAMA,
February
17,
1999—Vol
281,
No.
7
28. Physician-‐targeted
cost
reduc'on
efforts
• Speaker’s
preroga've
–
High
level
summary:
-‐ A
reasonable
number
studies
of
physician-‐
targeted
interven'ons
cost
and
u'liza'on
exist.
-‐ Example:
-‐ Provision
of
cost
informa'on
for
common
primary
care
medica'ons
increased
the
likelihood
that
lower
cost
alterna'ves
would
be
chosen
(Frazier
LM,
Ann
Intern
Med
1991;115:116-‐21.)
-‐ Recent
ar'cle
suggests
we
add
a
‘check
out’
cart
for
what
we
order
(Brook,
JAMA
2012)
29. Physician-‐targeted
cost
reduc'on
efforts
-‐ In
general:
-‐ Educa'onal
in
nature
-‐ Slight
decrease
in
u'liza'on
paKerns,
effect
did
not
differ
whether
IT-‐based
or
not
-‐ No
informa'on
as
to
whether
appropriateness
increased
-‐ Limited
persistence
of
interven'on
effect
-‐ Focus
on
variable
costs
(e.g.
lab,
some
drug
tests)
30. Health
systems
• Health
Technology
Assessment
CommiKees
• P&T
• An'microbial
stewardship
• Pathways
31. Health
Technology
Assessment
CommiKees
• Similar
to
pharmacy
and
therapeu'cs
commiKee
– In
existence
for
at
least
20
years
– Limited
data
on
their
prevalence,
but
appear
most
common
in
integrated
health
systems
• General
characteris'cs
– Broad
based
membership,
includes
C-‐suite
– Physician
led
and
championed
– Most
commonly
focus
on
surgical
technologies,
capital
expenditures
–
not
implants,
etc.
Fine
A
Healthc
Financ
Manage.
2003
May;57(5):84-‐7
32. Health
Technology
Assessment
CommiKees
• Few
(?No?)
data
on
their
effec'veness
in
constraining
costs
– In
general,
capital/technology
expenditures
represent
marketplace
differen'ators
and
are
hard
to
deny
• UCSF
HTAC
– In
existence
since
2006
– 24
approvals,
13
provisional
approvals,
4
declined
Gutowski
C
Health
Technology
Assessment
at
the
University
of
California–San
Francisco.
Journal
of
Healthcare
Management
56:1
January/February
2011
33. P&T
and
cost
• A
closed
formulary
may
produce
lower
pharmacy
costs
– Not
clear
whether
it
slows
pharmacy
cost
rises
– More
restric've
formulary
prac'ces
may
have
adverse
effects
(Horn
SD,
Formulary
limita'ons
and
the
elderly:
Results
from
the
Managed
Care
Outcomes
Project
AJMC
1198:
4;
1105-‐1113)
• No
data
on
whether
specific
P&T
structures
or
ac'vi'es
are
more
effec've
than
others
at
restraining
cost.
• No
data
on
the
‘return
on
investment’
of
P&T
34. An'microbial
stewardship
• Subtype
of
P&T
• Generally
narrower
focus
on
selected
set
of
medica'ons
• Able
to
link
choice
of
medica'ons
to
specific
clinical
situa'ons
(and
microbes)
– Oxen
include
clear
clinical
guidelines
– Pre-‐approval
via
consulta'on
in
many
programs
35. An'microbial
stewardship
• Highly
effec've
at
increasing
appropriateness
of
an'microbial
use
– Can
produce
both
reduc'ons
in
direct
costs
of
medica'ons
and
reduced
downstream
events
– Most
effec've
programs
include
physician
outreach/
approval
component
with
'ered
approach
• Limited
data
on
their
cost
to
benefit
ra'o
– One
study
es'mated
1M/year
direct
cost
savings
– No
es'mates
of
the
program
cost
Standiford
HC.
An'microbial
stewardship
at
a
large
ter'ary
care
academic
medical
center:
cost
analysis
before,
during,
and
axer
a
7-‐year
program.
Infect
Control
Hosp
Epidemiol
2012;33:338-‐45.
36. Health
systems
• An'microbial
stewardship
models
include
concepts
that
may
be
useful
elsewhere
– Physician
detailing
– Tiered
restric'on
process
– Understanding
that
restric'ng
an'microbials
has
real
benefits
• ASM’s
also
–
target
acute
care
medica'ons.
37. Pathways
• Where
do
guidelines/pathways
fit
in
this
talk?
– Lots
of
studies
on
guidelines,
pathways
– Minority
report
costs
as
an
outcome
– Few
used
guidelines/pathways
with
the
aim
of
reducing
costs/u'liza'on
(and
few
succeeded).
38. How
can
EMR’s
innovate
over
old
models
• HTAC,
P&T,
AMS,
and
Pathways
– Can
provide
decision
support
over
paper
models.
– Flexibly
catch
people
‘off
path’
– Set
firm(ish)
guardrails
around
unwanted
prac'ces
– Actually
measure
what
the
pathway
is
doing
and
give
clear
feedback
40. Ques'on
1:
Can
physicians
embrace
limita'ons
stewardship
as
a
professional
standard?
41. We
have
to
• Growing
recogni'on
that
costs
are
important
• EMR’s
can
help
reinforce
posi've
behavior,
discourage
unwanted
behavior
42. Ques'on
2:
Will
physicians
allow
themselves
to
be
stewarded
by
EMR’s
(or
anyone)?
43. The
pizalls
of
greater
choice
Physicians
want
this
The
future
needs
this
44. Find
‘Model
T’
systems
• What
would
your
company
look
like
if
it
produced
just
one
product?
– 1908
Ford
Model
T
– Single
model
from
which
others
arose
• More
modern
examples:
– Starbucks:
Medium
Coffee
– In
And
Out
Burger:
Cheeseburger
+
fries
– Bank
of
America:
Base
checking
account
45. Translate
this
to
healthcare
workflows
• Poten'al
examples
of
Model
T
workflows
– Inpa'ent
Model
T:
What
if
all
we
did
as
an
ins'tu'on
was
treat
pneumonia?
– Ambulatory
Model
T:
What
if
all
we
did
was
see
pa'ents
for
hypertension?
• Use
these
models
to
define
workflows,
clinic
visit
structure,
etc.
for
most
common
pa'ent
flows
– Add
‘modules’
on
to
allow
customiza'on
around
core
func'ons.
– Can
then
create
single
entry
points
an
economies
of
scale
– Can
EMR’s
do
this?
46. Translate
to
therapeu'c
choices
• Complexity
in
treatment
choices
– Within
these
groups,
what
20%
of
drugs
account
for
80%
of
pa'ents
• What
20%
of
devices
account
for
80%
of
pa'ents?
• Can
we
make
those
the
defaults?
• Can
this
approach
the
be
used
to
set
'ers
of
‘approval’
or
decision
support?
– Top
80%
easily
available
– Next
10%
require
aKesta'on
as
to
choice
– Next
10%
require
more
in-‐depth
approval?
47. Ques'on
3:
Can
health
IT
really
solve
everything
(including
costs)?
48. Health
informa'on
systems
and
cost
reduc'on
• HIT
is
thought
to
be
a
key
route
reducing
costs
– Reduced
redundancy
through
sharing
of
informa'on
– Ability
to
provide
decision
support
– Few
studies
have
demonstrated
cost
reduc'ons*
• Maybe
cost
increases??
49. Health
IT
• Why
it
might
be
old
wine
in
new
boKles
– Using
Health
IT
as….
• A
more
efficient
way
to
warehouse
as
broad
a
selec'on
of
treatments/algorithms
as
possible.
• A
more
efficient
way
to
present
all
possible
treatment/
tes'ng
choices
– We
don’t
improve
on
exis'ng
systems
and
organiza'onal
structures
to
help
define
appropriate
choices
• P&T,
An'microbial
“Culture
eats
technology
for
lunch”
50. Health
IT
• Why
it
can
get
us
to
the
place
we
want
to
be
– IT
does
force
workflow
standardiza'on
– More
sophis'cated
use
of
formulary
materials
limits
• Decision
support
• Automated
‘academic
detailing’
– Data
about
produc'on
processes
51. Conclusions
• Transla'ng
evidence
into
prac'ce
is
hard,
but
cri'cal
– EMR’s
can
help
by
providing
easy
access
to
the
best
therapies
and
tests
• Reducing
costs
in
EMR’s
is
a
cri'cal
goal
– Automate
and
smooth
workflows
to
target
fixed
costs
– Innovate
and
partner
with
exis'ng
organiza'onal
func'ons
to
speed
adop'on
52. Conclusions
• Can
we
find
answers
to
the
key
ques'ons?
– The
culture
of
healthcare
is
changing
– Perhaps
instead
of
culture
ea'ng
technology
for
lunch
(at
least
insofar
as
cost
and
quality
is
concerned),
culture
can
make
the
meal
more
appealing.