Structuring Recommendations for Clinical Decision Support
Slide Presentation from the AHRQ 2010 Annual Conference
On September 28, 2010, Jerry Osheroff made this presentation at the 2010 Annual Conference. Select to access the PowerPoint® presentation (3.6 MB). Free PowerPoint® Viewer (Plugin Software Help).
Structuring Care Recommendations for Clinical Decision Support
Jerry Osheroff, MD, Thomson Reuters
Jon White, MD, Agency for Healthcare Research and Quality
AHRQ Annual Meeting
September 28, 2010
Backdrop and Drivers for the eRecommendations Project
- AHRQ HIT Portfolio
- Federal CDS portfolio
- Federal CDS Collaboratory
- eRec Project Context: Improving Care through HIT
- eRec Project Overview:
- Engage stakeholders.
- Develop template, eRecs, and how-to guide.
- Vet deliverables for potential use.
- Next Steps:
- Engage more stakeholders.
- Vet and refine eRecs.
- Test drive eRecs.
eREC Project Context
Pressing Health Care Challenges
Cost and Efficiency
- Health spending =16% of GDP; > any other nation; $2.3 trillion; $7,600/person.
- Rising 6.9%/year (more than twice the inflation rate).
- 14% of U.S. population is uninsured.
- $700 Billion in waste.
Quality and Safety
- 44,000-98,000 preventable inpatient deaths/year.
- Patients have only 55% chance of appropriate care.
- Anticipate 17 years before effective treatment routine.
Sources: OECD Health Data, Thomson Reuters, Frost and Sullivan, IOM, Forbes, PwC Health Research Institute, Balas/IMIA, CITL, National Coalition on Healthcare
National Framework for Performance Improvement
Image: A flowchart depicts the National Framework. The first column is headed "Drivers and Reporting" and contains three steps:
- Incentivize outcome improvement and CDS use.
- Determine local performance improvement priorities.
- Gather and report provider performance data.
The second column contains three steps:
- Set National Improvement Priorities.
- Deploy Local Strategies to Drive Improvement (e.g. Decision Support).
- Realize Measurable Local/National Improvements and Repeat Cycle.
Arrows point down from each step to the one beneath it.
The third column is headed "CDS
Infrastructure" and contains two items:
- Best Practices for developing and using CDS to address goals.
- CIS/CDS tools available that help drive improvements.
Meaningful Use → Better Healthcare
"By focusing on 'meaningful use,' we recognize that better health care does not come solely from the adoption of technology itself, but through the exchange and use of health information to best inform clinical decisions at the point of care."
—David Blumenthal, 10/1/09
Government Role: HITECH Act
Image: Flowchart depicts the Government Role in the HITECH Act: Regional extension centers and Workforce training lead to Adoption of EHRs. Medicare and Medicaid incentives and penalties and Adoption of EHRs lead to Meaningful use of EHRs. State grants for health information exchange, Standards and certification framework, and Privacy and security framework all lead to Exchange of health information. Exchange of health information leads to Meaningful use of EHRs. Meaningful use of EHRs leads to Improved individual and population health outcomes, Increased transparency and efficiency, and Improved ability to study and improve care delivery. Research to enhance HIT is ongoing.
Blumenthal D. N Engl J Med 2009;10.1056/NEJMp0912825
HIT / EHR / CDS to the Rescue!
Outpatient: It's Just Not That Easy!
| Conclusion: As implemented, EHRs were not associated with better quality ambulatory care.
Electronic Health Record Use and the Quality of Ambulatory Care in the United States.
Jeffrey A. Linder, MD, MHP; Jun Ma, MD, RD, PhD; David W. Bates, MD, MSc; Blackford Middleton, MD, MHP, MSc; Randall S. Stafford, MD, PhD.
Arch Intern Med. 2007; 167 (13):1400-1405
Inpatient: Not Easy Here Either!
|Arch Intern Med 2005;165:1111-1116.
Original Investigation: High Rates of Adverse Drug Events in a High Computerized Hospital
Jonathan R. Nebeker, MS, MD; Jennifer M. Hoffman, PharmD; Charlene R. Weir, RN, PhD; Charles L. Bennett, MD, PhD, MPP; John F. Hurdle, MD, PhD
Advanced clinical systems with CDS
- 1/4 of admissions with at least 1 ADE; 9% serious harm.
- Problems with drug dosing, selection, monitoring.
What Do We Mean By CDS?
"...provide persons involved in care processes with general and person-specific information, intelligently filtered and organized, at appropriate times, to enhance health and health care"
- Includes and builds on current processes...
- Not just rules and alerts....
CDS Stakeholders Work in Relative Isolation on Very Difficult Problems
Image: A man underwater solving a Rubix cube is displayed.
Collaborative Effort on National CDS Strategy
Journal of the American Medical Informatics Association Volume 14 Number 2 Mar/Apr 2007
JAMIA Perspectives on Informatics
A Roadmap for National Action on Clinical Decision Support.
Jerome A. Osserhoff, MD; Jonathan M. Teich, MD, PhD; Blackford Middleton, MD, MPH, MSc; Elaine B. Steen, MA; Adam Wright; Don E. Detmer, MD, MA.
CDS Roadmap Pillars
Image depicting Enhanced Health & Healthcare Through CDS as a building; pillars consist of: Best Knowledge Available When Needed; High Adoption and Effective Use; and Continuous Improvement of CDS Methods and Knowledge. Best Knowledge Available When Needed points to Strategic Objective A.
|Strategic Objective A: Represent clinical knowledge and CDS interventions in standardized formats (both human and machine-interpretable), so that a variety of knowledge developers can produce this information in a way that knowledge users can readily understand, assess, and apply it.
Roadmaps for Successful CDS
Guidebooks on CDS Implementation for Providers:
- 2005 HIT book of the year
- All-time HIMSS bestseller
- Widely used by CMIOs/others
- 2011 Update in process
- Co-published by leading societies
- Over 100 contributors
- 2009 HIT book of the year
- Co-sponsors: AHRQ, 3 CIS vendors,.
- "This is not just a book"—ongoing collaboration
Notes: In parallel with the substantial national (and international) attention to the challenges associated with healthcare delivery, has been increasing attention to national efforts that can optimize the role of CDS in addressing these pressing challenges. One result from this attention is the Roadmap for National Action on CDS, presented to HHS Secretary Leavitt and AHIC in 2006. This Roadmap, developed by scores of stakeholders, has driven substantial public and private follow-up activities. Its key recommendations include developing better approaches to synthesizing and disseminating best practices for improving outcomes with CDS.
Toward this end, HIMSS published a CDS implementer's guide in 2005 that has been well received and widely used to improve the efficiency and effectiveness of provider organizations' CDS efforts. In follow-up to this successful guide, nearly 100 contributors from dozens of organizations have come together over the last few years to build on the framework from this previous work to offer recommendations on improving outcomes in a specific, high priority area—i.e. improving medication use and outcomes. The result of this highly collaborative effort has been published in early 2009 by HIMSS and many other leading societies, and co-sponsored by AHRQ, leading CIS vendors and others. The book is offered by the collaborative group that produced it not as a ‘finished product' but rather as a snapshot of an ongoing conversation among CDS implementers about how to best enhance the success and value of their efforts.
This slides that follow present concepts, recommendations and figures used in this latest CDS implementer's guide.
A Formula for Success: The CDS Five Rights
To improve care outcomes with CDS, you must provide:
- The Right Information: Evidence-based, useful for guiding action and answering questions.
- To the Right Stakeholder: Both clinicians and patients.
- In the Right Format: Alerts, order sets, answers, etc.
- Through the Right Channel: Internet, mobile devices, clinical information systems.
- At the Right Point in the Workflow: To influence key decisions/actions.
Some Sources for "The Right Information"
Image: A flowchart. The left side section reads, "Clinical and outcomes research (What works?)" which leads to a text box with: "What should be done? (CPGs)," which points to "What should be done now? (CDS)," which points to "How are we/am I doing? (Performance measurement/reporting)," which points back to the top text box: "What should be done? (CPGs)."
Getting the 'Right Information' Into Clinical Practice
In Theory, Straightforward:
- Evidence → Guidelines
- Guidelines → Changes in clinical practice
- Changes in practice → Improved quality of care
Getting the 'Right Information' Into Clinical Practice
In practice, barriers are widespread:
- Evidence basis has gaps and inconsistencies.
- Physicians disagree with guidelines or patients may not comply.
- Inertia exists; incentives to change are lacking; disincentives exist.
- Volume of guideline content is large and hard to track; accessible content at right time in care process is missing.
- Difficulty of implementing guidelines (in information systems):
- Guidelines have free-text format, ambiguous terminology, lack of data elements/data schema in published guidelines.
- Implementation is complex and site-specific (e.g., workflow).
eREC Project Overview
eREC Project Goal
- To accelerate widespread uptake of well-accepted, evidence-based patient care recommendations into clinical information systems:
- By developing a formal method for translating narrative into structured, coded logic statements.
- Useful for further local processing into CDS rules.
AHRQ eREC Project Team
- Thomson Reuters
- Project Director: Jerry Osheroff, MD
- Susan Raetzman, Rosanna Coffey, Andriana Hohlbauch and others
- Technical Lead: Robert Greenes, Arizona State University
- ERec Developer: Margarita Sordo, Mass Gen Hosp, Harvard Med
- Peter Haug, Intermountain Health Care
- Aziz Boxwala, University of California at San Diego
- Ted Shortliffe, American Medical Informatics Association
- Key Collaborators
- Jacob Reider, Electronic Health Records Association
- Floyd Eisenberg, National Quality Forum
- William Bria and select AMDIS members
eREC Project Activities
- Needs and prior work: Synthesize stakeholder needs and related efforts.
- ERec Format: Develop format for converting guideline recommendations into structured logic statements.
- ERecs Applied: Convert 47 recommendations into the structured logic format:
- 45 "A and B" USPSTF recommendations.
- 2 clinically relevant Meaningful Use criteria.
- Processes and lessons—so others can replicate and learn.
- Disseminate results—so CDS implementation accelerates.
Focus of Needs and Prior Work
|CDS vendors and implementers
||What will make eRec products most useful in process of translating guidelines into machine rules?
|Providers of care
||What works well or is problematic in CDS products and processes?
|Standards setting organizations
||How can existing standards be used in new format for translating care recommendations?
|Quality improvement organizations
||Can performance measurement momentum be leveraged? Can eMeasures inform eRecs?
||Can the development of care guidelines be improved/informed by using eRec format?
Needs and Prior Work: Findings
Build on knowledge-sharing collaboratives
- Translation is multi-step process.
- Other formalisms exist (HL7 RIM, GEM, etc.).
- Create a semi-structured formalism.
- Leverage other formalisms as appropriate.
eREC Project in CDS Context: Stage 2
Image: A flow chart:
|Stages of Rule
|1. Free-text logic statement
||[Arrow pointing left] Assemble Knowledge
- Assemble elements of narrative guideline needed to produce a logical statement
- Include other CDS-related elements
|2. Structured logic statement
||[Arrow pointing left] Create Structured Logic Statement
- Express medical knowledge in structured format that codifies data and logical expressions
- Flag and annotate items that require further disambiguation
- Identify key implementation considerations
|3. Pre-executable logic statement
||[Arrow pointing left] Translate Statement to Pre-executable Format
- Evaluate logic statement in use scenarios
- Incorporate attributes that anticipate local implementation considerations, data types, and rule triggering scenarios
|4. Deployable logic statement
||[Arrow pointing left] Generate Deployable Rules
- Develop setting-specific representations for local systems
- Ensure the rule can be engineered into HIS and care setting
Needs and Prior Work: Findings (cont'd)
Common needs vs. setting specific needs
- Substantial effort and duplication for translation.
- Clinical assumptions are not always explicit.
- Implementers want disambiguated logic statements and clearly defined and coded data elements.
- Workflow considerations are highly local; tension over specificity in addressing these.
- Provide data definitions and codes where possible.
- Include "Implementation Considerations": Less specificity of workflow considerations in logic increases portability and allows local tailoring.
Importance of the Local Context for Applying CDS
Flowchart showing the steps in applying CDS in a local context:
Step 1: Establish CDS Management Charter, Governance; Engage Stakeholders which leads to: Step 2: Determine Opportunities, Goals, Baselines which leads to: Step 3: Examine Workflows, Infrastructure which leads to: Step 4: Configure Internventions to Test Goals which goes down to: Step 5: Test Interventions; Communicate, Train, Launch which leads to Step 6: Assess/Improve. An arrow then goes back toStep 2: Determine Opportunities, Goals, Baselines.
There is also a text box in the middle which states: Manage CDS Assets, Decisions, Processes
Figure ©2009 HIMSS
Needs and Prior Work: Findings (cont'd)
Quality improvement efforts:
- National push for Meaningful Use of HIT.
- Health Quality Measures Format (HQMF): standard for expressing quality measure in format for EHR integration, i.e., eMeasure .
- NQF Quality Data Set (QDS): Common language for information in quality measures, e.g., data elements, code lists, care setting attributes.
- Desirable to leverage momentum and related tools.
- ERecs related in concept and content to HQMF and eMeasure.
- Some adjustments needed: Performance measures are population based; CDS based on patient-provider encounter.
eREC Project Conceptual Approach
Leveraging Quality Measurement Standards and EHR Integration to Support Widely Useful Structured Recommendations for CDS Rules
Evidence-Based Care Guidelines, e.g.:
- USPSTF A&B-graded recommendations
- Interventions underlying meaningful use measures
CDS Interventions: eRecommendations
- eRecommendation operational exclusion criteria
- Other CDS implementation considerations
Clinical Information Systems
- eRecommendation eligibility criteria/eMeasure denominator criteria
- Exclusion criteria
- Action recommended/action measured
Quality Measures: eMeasures (in HQMF Format)
Value Sets, Code Sets, Code Lists, Quality Data Types:
- Unfolding work of NQF, HITSC, etc.
- Three main parts to eRecommendation format
- Header—information describing eRec and underlying clinical care recommendations.
- Data Definition and Logic Specification—identifies data elements, code sets, and values needed to express logic; provides logic statement for identifying patients who satisfy criteria for care recommendation.
- Implementation Considerations—lists other issues that care providers and vendors should consider when implementing for local settings.
eREC Format: Header Section
Image: Screenshot of Header section from eRec format spreadsheet.
eREC Format: Data and Logic Section
Image: Screenshots of Data and Logic Specification section of spreadsheet and Logic Statement section of spreadsheet.
eREC Format: Implementation Considerations Section
Image: Screenshot of Implementation Considerations section of spreadsheet.
eREC Dissemination Products Available from Project
- Methods Report: Background, existing approaches, approach for eRecommendations.
- eRec Template: Format for developers, vendors, implementers.
- eRecs of two types:
- 45 A- and B-graded recommendations from the USPSTF.
- 2 Stage 1 Meaningful Use criteria.
- Available on AHRQ/NRC site when final. For Excel example of future eRec, E-mail email@example.com.
- Standard Operating Procedures (SOPs): How to apply eRec template to care recommendations.
eREC Project Impact (To Date)
- Stimulating broad conversation among key CDS players (guideline suppliers, CDS implementers).
- Cultivating synergies between CDS and performance measurement (from goals to codes).
- Garnering attention of guideline developers.
- Illustrating the concept of formal logic structures to support measurable, CDS-enabled healthcare performance improvement.
Next Steps (proposed)
- Pilot eRecs in real world settings (EP/EH):
- Focus on MU clinical topics.
- Flesh out implementation considerations.
- Build 'value chain community' to follow and help drive to scale:
- Guideline suppliers, CIS suppliers, implementers, federal stakeholders, etc.
- Develop eRecs for additional MU measures, based on implementer need.
Vision Beyond Project
- eRec as standard for expressing guidelines.
- Key guideline developers produce guidelines in eRec format for quick uptake into CDS.
- CIS vendors use eRecs as part of CDS capabilities deployment.
- Care delivery organizations implementing CDS adopt guidelines rapidly.
- Gain insights on and improve guidelines-to-alerts-to-better-outcomes chain of events.
- eRecs help drive measurable care improvements.
Thank You For Your Interest!
For more information:
Current as of November 2010
Structuring Recommendations for Clinical Decision Support. Slide Presentation from the AHRQ 2010 Annual Conference (Text Version). November 2010. Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/about/annualconf10/osheroff/osheroff.htm