Transcript Shiffman_2011

NYAM Teach 2011
Making Guidelines WORK
Richard N. Shiffman, MD, MCIS
Yale School of Medicine
Disclosure
• Funding from AHRQ and NLM
• Will describe several applications
developed in our laboratory
available gratis for non-commercial
use
Today
• Making Guidelines THAT Work
– Improving the product
– Clarity, transparency, and implementability
– GLIA and BRIDGE-Wiz
• Making Guidelines WORK
– Improving the delivery of knowledge to the point
of care
– Computer-based clinical decision support
3
Guidelines have problems…
• Cluzeau (Int J Qual Healthcare 1999), Shaneyfelt (JAMA
1999) majority of guidelines failed quality criteria
• Grilli: 431 specialty society guidelines (Lancet 2000)
– 82% did not apply explicit criteria to grade evidence
– 87% did not report whether a literature search was
performed
– 67% did not describe type of professionals involved in
development
• Shaneyfelt (JAMA 2009): persisting biases; lack of
specificity, flexibility, regular updating
• Alonso-Coello: in 42 reviews of 626 guidelines over past 20
years, mean quality scores for rigor of development,
stakeholder involvement, editorial independence, and
applicability are “moderate” or “low” (GIN 2009)
CPGs are statements
that include
recommendations
intended to optimize
patient care that are
informed by a
systematic review of
evidence and an
assessment of the
benefits and harms of
alternative care options
Guidance on the use of glitazones for the
treatment of type 2 diabetes
• For people with type 2 diabetes, the use of a glitazone
as second-line therapy added to either metformin or a
sulphonylurea--as an alternative to treatment with a
combination of metformin and a sulphonylurea-- is
not recommended except for those who are unable to
take metformin and a sulphonylurea in combination
because of intolerance or a contraindication to one of
the drugs. In this instance, the glitazone should
replace in the combination the drug that is poorly
tolerated or contraindicated.
Authors Should Be Explicit About
•
•
•
•
•
•
•
IF
Denominator
WHEN {under what circumstances}
WHO {in the Intended Audience}
THEN
Numerator
Ought to {with what level of obligation}
DO WHAT
{To WHOM} {which members of the target population}
HOW
WHY
Guidance on the use of glitazones
for the treatment of type 2 diabetes
• If a patient is unable to take the combination of
metformin and sulfonylurea (because of
intolerance or contraindication), the clinician
should prescribe a glitazone to replace the drug
that is not tolerated.
Guidance on the use of glitazones
for the treatment of type 2 diabetes
UNDER WHAT CIRCUMSTANCES?
• If a patient is unable to take the combination of
metformin and sulfonylurea (because of
intolerance or contraindication), the clinician WHO?
should prescribe a glitazone to replace the drug
that is not tolerated.
OUGHT?
To do WHAT?
Statement of fact is NOT
a recommendation
• Adjuvant hormone therapy for locally
advanced breast cancer results in
improved survival in the long term.
• Clinicians should prescribe adjuvant
hormone therapy for locally advanced
breast cancer (when/unless?)…
How “Should” We Write
Guideline Recommendations:
Interpretation of Deontic Terminology
Lomotan E, et al. Qual & Safety in Health Care 2010
• Goal: To describe the level of obligation
conveyed by deontic terms commonly used in
practice guidelines
• Can level of obligation be standardized?
Measuring Obligation
0
50
100
Level of Obligation
100
75
50
25
0
Musts (19/1250 – 1.5%)
• Narcotic use must be carefully titrated and supervised.
• Clinicians working in juvenile justice settings must be vigilant for
personal safety and security issues and aware of actions that may
compromise their safety and/or the safety and containment of the
incarcerated youth
• Nurses working with individuals with asthma must have the
appropriate knowledge and skills to identify the level of asthma
control, provide basic asthma education, conduct appropriate
referrals to physician and community resources
• Treatment of duodenal adenomas depends on adenoma size and
the presence of severe dysplasia. Small tubular adenomas with mild
dysplasia can be kept under surveillance, but adenomas with severe
dysplasia must be removed
15
The Dreaded “Consider”
• The Expert Panel concludes that initiating daily longterm control therapy should be considered for
reducing impairment in infants and young children
who consistently require symptomatic treatment more
than 2 days per week for a period of more than 4
weeks (Evidence D).
• Referral may be considered if a child 0–4 years of
age requires step 2 care or a child 5–11 years of age
requires step 3 care.
Measurement
• If you can’t measure it, you can’t manage it.
Peter Drucker
• If you don’t measure it, you can’t improve it.
Action-Types
Gather Data
Interpret
Test
Conclude
Inquire
Monitor
Examine
Act
Prescribe
Educate/counsel
Document
Prevent
Procedure
Consult/refer
Advocate
Prepare
Dispose
Action-Type Pattern: Prescribe
• Drug information
• Safety alerts (allergy, drug-drug, drugdisease, drug-lab)
• Formulary check
• Dosage calculation
• Pharmacy transmission
• Patient education
• Corollary orders
A Transparent Process
for Generating Recommendations
A transparent development process
makes clear…
• How authors weighed
• evidence
• pathophysiologic reasoning (first principles)
• expert experience
• patients’ and society’s values
• Allows users to judge reasonableness of
recommendations
Requires untangling and specifying
2 related (but distinct) concepts
{Elegant and erudite work of GRADE Collaboration}
• Quality of evidence
<---developers’
focus
• Recommendation strength<-what implementers need to know
to design systems that influence care
• level of expected adherence
• level of enforcement / incentive
Evidence Quality
• An indication of the authors’ confidence in
their appraisal of benefits and harms
• Based on an analysis of the validity,
consistency, and directness of the evidence
supporting a recommendation
Recommendation Strength
• Implementers need to understand experts’
assessment of strength of recommendation
• Communicates authors’ assessment of the importance
of adherence
• Levels based on aggregate evidence quality and
balance of anticipated benefits and harms
– Strong recommendation (“MUST”)
– Recommendation (“SHOULD”)
– Option (“MAY”)
Grading Recommendation Strength
Evidence Quality
A. Well designed RCTs or diagnostic studies
on relevant population
Preponderance
of Benefit or
Harm
Strong
B. RCTs or diagnostic studies with minor
limitations;overwhelmingly consistent
evidence from observational studies
C. Observational studies (case-control and
cohort design)
D. Expert opinion, case reports, reasoning
from first principles
X. Exceptional situations where validating
studies cannot be performed and there is a
clear preponderance of benefit or harm
Balance of
Benefit and
Harm
Rec
Option
Strong
Rec
Option
No Rec
GuideLine
Implementability
Appraisal
BMC Medical Informatics and Decision Making
2005
• Goals
– To identify intrinsic obstacles to implementation, i.e., those
that are within the purview of guideline developers
– To provide feedback to guideline authors to anticipate and
address these obstacles before a draft guideline is finalized
– To assist implementers in guideline selection and to target
attention toward anticipated obstacles
• GLIA (and eGLIA) available from http://gem.med.yale.edu/glia
GLIA v2.0 Dimensions
• Decidability - precisely under what conditions (e.g., age,
gender, clinical findings, lab results) to do something
• Executability - exactly what to do under the circumstances
defined)
• Validity - the degree to which a recommendation reflects
the intent of the developer and the strength of evidence
• Flexibility - the degree to which a recommendation permits
interpretation and allows for alternatives in its execution
• Effect on process of care - the degree to which a
recommendation impacts upon the usual workflow in a
typical care setting
GLIA v 2.0 Dimensions (cont’d)
•
•
•
Measurability – the degree to which the guideline
identifies markers or endpoints to track the effects of
implementation of this recommendation
Novelty/innovation - the degree to which a
recommendation proposes behaviors considered
unconventional by clinicians or patients
Computability - the ease with which a recommendation
can be operationalized in an electronic information system
Bridge the Gap
Between Authors and Implementers
With BRIDGE-Wiz
(Building Recommendations In a Developer’s Guideline Editor)
BRIDGE-Wiz
• Displays a sequence of screens representing chunks of
information about a recommendation
• The authors systematically and sequentially determine:
–
–
–
–
action(s) to be recommended
condition(s) under which the action is to be performed
benefits, risks, harms, and costs of the proposed action
the quality of the evidence supporting the action.
• The program’s output is an IF…THEN rule and
supporting recommendation profile
30
Bridge-Wiz Demo
31
BRIDGE-Wiz
Building Recommendations in a Developer’s Guideline Editor
•
•
•
•
Formalizes a process for writing implementable recommendations
Focuses discussion
Incorporates prompts based on COGS to improve guideline quality
Controlled natural language
– Offers verb choices based on action-type
– Traps and disallows use of “consider”
– Discourages “statement of fact” masquerading as recommendation
– Limits boolean connectors to all ANDs or ORs in a statement
• Incorporates decidability and executability checks
• Requires systematic appraisal of evidence quality and benefit-harms
– Suggests appropriate obligation term (deontic modal)
• Output includes a high-level “rule” and an evidence profile
Making Guidelines Work
33
Interventions to Influence Practice
Grol, Grimshaw Lancet 2003
•
•
•
•
•
Education (conferences, courses)
Audit & feedback
Financial incentives/disincentives
Patient-mediated interventions
Computer based decision support
34
Clinical Decision Support: Definition
• Use of the computer to bring relevant knowledge to bear
on the health care and well-being of a patient (Greenes).
• Systems that link health observations with health
knowledge to influence health choices by clinicians for
improved health care (Hayward)
35
Computer-Based Decision Support
Systematic Reviews
Mary Johnston
Derek Hunt
Amit Garg
Ken Kawamoto
Basit Chaudhry
McMaster
McMaster
Univ. Western
Ontario
Duke
UCLA
JAMA 1994
JAMA 1998
JAMA 2005
BMJ 2005
Ann Intern Med
2006
•Computer-based decision support regularly—but not
always—improves the process of care
•Outcomes—though infrequently measured—sometimes
improve
Identifying Features Critical to Success
Kawamoto K. BMJ 2005
• Significant improvement in practice in 68% of 70 trials
• Predictors of improved practice:
– Automatic provision of DS as part of workflow
– Providing DS at time and site of decision making
– Providing recommendations, not just assessments
– Providing periodic performance feedback
– Sharing recommendations with patients
– Requesting reasons for not following recommendations
Allergy Alert
38
Palette of CDS Interventions
Alert
Documentation
Reminder
template
Algorithm
Order
Facilitator
Flowsheet
Infobutton Calculator
Selected Guideline
•
Asthma
– EPR3 Diagnosis and Management of Asthma from
the NHLBI (2007)
– Demonstrates challenges involved in implementation
of recommendations for chronic management of
complex disease
40
Prompts for documentation
Real-time calculation and display
Display of Relevant
Prompts for
Assessments Past Information
Alert
Information Access
Order Set
Customizable Handout
Medication Authorization
46
Summary
• Making Guidelines THAT Work
– Must address:
– Clarity, transparency, and implementability
– GLIA and BRIDGE-Wiz
• Making Guidelines WORK
– Computer-based clinical decision support
– Improving the delivery of knowledge to the point
of care
47
Thank You!
ycmi.med.yale.edu/GLIDES
[email protected]