Kensaku Kawamoto, MD, PhD Initiative Coordinator, Health eDecisions Associate Chief Medical...
-
Upload
christiana-ivy -
Category
Documents
-
view
221 -
download
2
Transcript of Kensaku Kawamoto, MD, PhD Initiative Coordinator, Health eDecisions Associate Chief Medical...
Kensaku Kawamoto, MD, PhDInitiative Coordinator, Health eDecisions
Associate Chief Medical Information Officer, Univ. of Utah Health Sciences Center
Tonya Hongsermeier, MD, MBAChief Medical Information Officer, Lahey Health
Aziz Boxwala, MD, PhD, FACMIPresident, Meliorix, Inc.
Victor C. Lee, MDVice President of Clinical Informatics, Zynx Health
Jacob Reider, MDActing National Coordinator for Health IT
Health eDecisions: A Public-Private Partnership to Enable Standards-
Based Clinical Decision Support at Scale
1
Introductions and Lineup
Panelist Title Initiative Role
Jacob Reider, MD Acting National Coordinator for Health IT
Executive Sponsor
Tonya Hongsermeier, MD, MBA
CMIOLahey Health
Past Initiative Coordinator
Kensaku Kawamoto, MD, PhD
Associate CMIO University of Utah
Initiative Coordinator
Aziz Boxwala, MD, PhD
PresidentMeliorix, Inc.
Subject Matter Expert
Victor Lee, MD VP of Clinical InformaticsZynx Health
Key Community Contributor
2
Agenda
Topic Panelist
ONC perspective on motivation for initiating HeD Reider
Overview of HeD and its methodology Hongsermeier
Foundational data model: HL7 Virtual Medical Record Kawamoto
HeD Use Case 1: sharing CDS through standard knowledge artifacts
Boxwala
HeD Use Case 2: sharing CDS through CDS guidance services
Kawamoto
Pilots overview and vendor perspective Lee
Panel discussion, and questions from audience Reider
3
Jacob Reider, MDActing National Coordinator for Health IT
Executive Sponsor, Health eDecisions
ONC Perspective on Motivation for Initiating HeD
4
The Promise of CDS
• RCTs demonstrating CDS effectiveness available for almost 40 years1
• Actionable, computer-generated CDS provided automatically at the point of care has significantly improved care quality in >90% of RCTs2
Ref 1. McDonald C. NEJM. 1976;295:1351-5.Ref 2. Kawamoto K et al. BMJ. 2005;330:765.
5
The Reality of CDS
• CDS commonly available for drug-drug interactions and drug-allergy contraindications
• Robust CDS for other domains available in select healthcare systems
• However, in general, CDS use is relatively limited
Ref: Osheroff JA et al. A roadmap for national action on clinical decision support. J Am Med Inform Assoc. 2007;14:141-145.
6
Key Barrier: Limited CDS Portability
• Most existing CDS systems and their knowledge bases have limited portability1
Ref 1: Osheroff JA et al. J Am Med Inform Assoc. 2007;14:141-145.
7
Potential Solutions
• Define standard, universal format(s) for CDS knowledge that can be written once and imported anywhere
• Define standard, universal format for encapsulating and accessing CDS capabilities as a software service
8
Why HeD?
• “The nice thing about standards is that there are so many to choose from”– Andrew Tanenbaum
• Relevant prior work that can be leveraged, plus need to align and harmonize
• Market forces alone maybe inadequate to move CDS-enabled health and healthcare improvement forward
9
The Opportunity: Alignment with MU• Meaningful Use (MU)
– Federal program that incentives adoption of EHR systems fulfilling MU certification requirements
– Administered by U.S. Office of the National Coordinator for Health IT (ONC)
• MU Focus– Stage I (2011-2012): data capture and sharing– Stage II (2014): advance clinical processes– Stage III (2016): improved outcomes, with CDS as a key
component
10
Tonya Hongsermeier, MD, MBAChief Medical Information Officer, Lahey Health
Past Initiative Coordinator, Health eDecisions (Jun-Dec 2012)
Overview of HeD and its Methodology
11
Health eDecisions (HeD)
• ONC-sponsored, public-private initiative to develop and validate standards to enable CDS at scale (www.healthedecisions.org)– Will inform MU Stage III EHR certification criteria
• Brief timeline– April 2012: face-to-face planning discussion, DC– June 2012: initiative kickoff– Jan, May, Sept 2013: HL7 ballots– March 2013+: pilots– Late ’13/early ’14 (anticipated): draft MU criteria
12
Key Contributors• Wide, deep range of contributors• ONC: Jacob Reider, Alicia Morton, Joe Bormel, Amy Helwig• Initiative Coordinator: Ken Kawamoto (1/13+), Tonya
Hongsermeier (6/12-12/12)• Initiative SMEs: Aziz Boxwala, Bryn Rhodes• Terminology SMEs: Robert McClure, Mark Roche• Project Management and Support: Jamie Parker, Atanu Sen, Anna
Langhans, Saunya Williams, Virginia Riehl, Divya Raghavachari• Community Contributors: Zynx Health (Claude Nanjo* and Victor
Lee*), Univ. of Utah (David Shields*), Intermountain Healthcare, ASU, Wolters Kluwer Health, VHA, Allscripts, newMentor, CDC, Design Clinicals, and many, many others
*Each with hundreds of hours of contribution 13
HeD Goal
• To define and validate standards that facilitate the emergence of systems and services whereby CDS interventions can be shared or accessed by any healthcare stakeholder via an importable format or via a CDS Web service
• In short, to define and validate standards that enable CDS sharing at scale
14
HeD Scope
• Use Case 1: standard format for sharing CDS knowledge artifacts– Rules, order sets and documentation templates
– Goal: CDS knowledge authored in standard format can be imported and used in any EHR system
• Use Case 2: standard interface for accessing CDS Web services– Goal: CDS capability encapsulated using standard
interface can be integrated with any EHR system
15
Standards & Interoperability (S&I) Framework Methodology
• Structured framework for defining and validating standards for consideration in MU
Tools and ServicesTools and Services
Use Case Development
and Functional Requirements
Use Case Development
and Functional Requirements
Standards DevelopmentSupport
Standards DevelopmentSupport
Harmonization ofCore Concepts
Harmonization ofCore Concepts
Implementation Specifications
Implementation Specifications
Pilot Demonstration Projects
Pilot Demonstration Projects
Reference Implementation
Reference Implementation
Architecture Refinement and ManagementArchitecture Refinement and Management
Certificationand TestingCertificationand Testing
16
Prior Work Analyzed• Standard, universal format for CDS knowledge
– HL7 Arden Syntax, HL7 GELLO, HL7 Order Sets, ASTM GEM, GLIF3, CDS Consortium L3 model, SAGE, Asbru, PROforma, PRODIGY, AHRQ eRecommendations, etc.
• Standard interface for submitting patient data and obtaining patient-specific care guidance– HL7 Decision Support Service, IHE Request for Clinical
Guidance, OpenCDS, CDS Consortium ECRS, SEBASTIAN, etc.
• Standard information models– HL7 Virtual Medical Record, C-CDA, QRDA, various HL7
V3 models, FHIM, FHIR, etc.
Ref. Kawamoto K et al. Open Medical Informatics Journal. 2010, 4:235-44. 17
Health eDecisions TimelineMarch MayFeb Apr June July Aug Dec
Consensus Reached on UC 2
UC Harmonization Begins
Pilot Work Begins Pilot Activities Completed
18
June 12
NovSept Aug Dec Jan 13
Kickoff
Consensus on Project Charter Reached
Consensus Reached
Consensus Standards Reached
HL7 Ballot Submitted
Presented at HL7 – Affirmative Vote Received
Pre-Discovery
Discovery Implementation Pilot
Use
Cas
e O
ne
Use
Cas
e Tw
o
Discovery Implementation
Present 5 UC 2 Artifacts to Sept Hl7 Ballot
Sept Oct Nov
Reconcile all ballot comments and submit UC 2 Artifacts to HL7 for Jan 2014 Ballot
18
Deliverables Overview
• Foundational Deliverables– Functional requirements, including scope specification,
use cases, and data requirements– Analysis of relevant efforts (esp. standards)
• Standards– HL7 Virtual Medical Record (multiple facets)– HL7 Decision Support Service Release 2– Use Case 1 and 2 implementation guides
19
Kensaku Kawamoto, MD, PhDAssociate Chief Medical Information Officer, University of Utah
Initiative Coordinator, Health eDecisions
HeD Key Deliverable 1: Foundational CDS Information Model
20
Disclaimers• I am, or have been in the recent past, a consultant on
clinical decision support to the following entities:
• I have no competing interests related to OpenCDS
21
Office of the National Coordinator for Health IT (ONC)
Partners HealthCare RAND Corporation ARUP Laboratories
McKesson InterQual ESAC, Inc. Inflexxion, Inc. Intelligent Automation, Inc.
Underlying Information Model
• Need– Standard CDS data model that is simple and intuitive for a
typical CDS knowledge engineer to understand and use
• Relevant Prior Work Evaluated– HL7 Consolidated Clinical Document Architecture (C-CDA)– HL7 Quality Reporting Document Architecture (QRDA)– HL7 Fast Healthcare Interoperability Resources (FHIR)– HL7 Virtual Medical Record (vMR)– IHC Clinical Element Models, OpenEHR templates, others
• Decision– HL7 vMR with templates derived from C-CDA and QRDA
22
vMR Background
• A “holy grail” of clinical informatics is scalable, interoperable CDS
• Key requirement for interoperable CDS and re-use of CDS knowledge resources = use of a common patient data model– Referred to as a “Virtual Medical Record” or vMR
(Johnson et al., AMIA Annu Symp Proc, 2001)
• Needs to be easy and safe for a typical CDS knowledge engineer to understand and use
• Lack of a common vMR has been a major barrier to sharing knowledge and scaling CDS
23
Example Challenge without VMR
ObservationCode = BPValue = 120/80 mmHg
Blood PressureSystolic = 120 mmHgDiastolic = 80 mmHgCode = BPValue = 120/80 mmHg
Observation Code = BP
ObservationCode =
SBPValue =
120 mmHGObservation
Code = DSP
Value = 80 mmHg
Vital SignsType = BPValue = 120/80Units = mmHg
24
vMR Goal
• Provide standard information model for CDS that (i) can be used across CDS implementations and (ii) is simple and intuitive for a typical CDS knowledge engineer to understand, use, and implement
25
Development Methodology
• Analysis of data required by 20 CDS systems from 4 countries (Kawamoto et al., AMIA 2010)
• Analysis of various standard information models– HL7 CCD, C-CDA, QRDA, Pedigree model, Clinical
Statement pattern, etc.
• Analysis of orderables from hundreds of hospitals• Iterative refinement from pilot use
– In particular, through OpenCDS (www.opencds.org) and HeD pilots
– Initial ballot in 2010; Sept. 2013 = 4th round of balloting
26
Why Not Just Use C-CDA as the vMR?
<entry typeCode="DRIV"> <act classCode="ACT" moodCode="EVN"> <templateId root="2.16.840.1.113883.10.20.22.4.3"/> <id root="ec8a6ff8-ed4b-4f7e-82c3-e98e58b45de7"/> <code code="CONC" codeSystem="2.16.840.1.113883.5.6" displayName="Concern"/> <statusCode code="completed"/> <effectiveTime><low value="20070103"/></effectiveTime> <entryRelationship typeCode="SUBJ"> <observation classCode="OBS" moodCode="EVN">
<templateId root="2.16.840.1.113883.10.20.22.4.4"/> <id root="ab1791b0-5c71-11db-b0de-0800200c9a66"/> <code code="409586006" codeSystem="2.16.840.1.113883.6.96" displayName="Complaint"/> <statusCode code="completed"/> <effectiveTime><low value="19500101"/></effectiveTime> <value xsi:type="CD" code="195967001" codeSystem="2.16.840.1.113883.6.96" displayName="Asthma"/> <entryRelationship typeCode="REFR"> <observation classCode="OBS" moodCode="EVN">
<templateId root="2.16.840.1.113883.10.20.22.4.6"/> <code xsi:type="CE" code="33999-4" codeSystem="2.16.840.1.113883.6.1" codeSystemName="LOINC" displayName="Status"/> <statusCode code="completed"/> <value xsi:type="CD" code="55561003" codeSystem="2.16.840.1.113883.6.96" codeSystemName="SNOMED CT" displayName="Active"/> </observation> </entryRelationship> </observation> </entryRelationship> </act> </entry>
CCDA 1.1 representation of “Patient has had asthma since 1950”
27
Why Not Just Use C-CDA as the vMR?
entry[typeCode=“DRIV” and act[classCode=“ACT” and moodCode=“EVN” and /templateId[root=“2.16.840.1.113883.10.20.22.4.3”] and /code[codeSystem=“2.16.840.1.113883.5.6” and code=“CONC”] and /statusCode[code=“completed”] and /entryRelationship[typeCode=“SUBJ” and /observation[classCode=“OBS” and moodCode=“EVN” and /templateId[root=“2.16.840.1.113883.10.20.22.4.4”] and /code[codeSystem=“2.16.840.1.113883.6.96” and code=“409586006”] and /statusCode[code=“completed”] and /effectiveTime[/low[value<=“20130814”]] and
/value[xsi:type=“CD” and codeSystem=“2.16.840.1.113883.6.96” and code=“95967001”] and /entryRelationship[typeCode=“REFR” and /observation[classCode=“OBS” and moodCode=“EVN” and /templateId[root=“2.16.840.1.113883.10.20.22.4.6”] and /code[xsi:type=“CE” and codeSystem=“2.16.840.1.113883.6.1” and code=“33999-4” ] and /statusCode[code=“completed”] and /value [xsi:type=“CD” and codeSystem=“2.16.840.1.113883.6.96” and code=“55561003” ] ] ] ] ] ]
Sample CDS expression that“Patient currently has active asthma” using CCDA 1.1 Data Model
28
vMR Representation of Equivalent Content
<clinicalStatement xsi:type="vmr:Problem"> <templateId root="2.16.840.1.113883.3.1829.11.7.2.5"/> <problemCode codeSystem="2.16.840.1.113883.6.96" code="195967001"><displayName value="Asthma"/></problemCode> <problemEffectiveTime><low value="19500101"/><problemEffectiveTime> <problemStatus codeSystem="2.16.840.1.113883.6.96" code="55561003"><displayName value="Active"/></problemStatus></clinicalStatement>
Sample CDS expression for “Patient currently has active asthma” using vMR
clinicalstatement[xsi:type=“vmr:Problem” and /templateId[root=“2.16.840.1.113883.3.1829.11.7.2.5”] and /problemCode[codeSystem=“2.16.840.1.113883.6.96” and code=“195967001”] and /problemEffectiveTime[/low[value<=“20130814”]] and /problemStatus[codeSystem=“2.16.840.1.113883.6.96” and code=“55561003”] ]
vMR representation of “Patient has had asthma since 1950”
29
Simplification – Data Types
HL7 Version 3 Release 2 Data Type Model for Integer
Constrained HL7 Version 3 Release 2 Data Type Model for
Integer used in vMR
Value of the INT
30
Example vMR Template
Example:<clinicalStatement xsi:type="vmr:Problem"> <templateId root="2.16.840.1.113883.3.1829.11.7.2.5“ identifierName=“ActiveProblemListEntryCodeOnly”/> <problemCode codeSystem="2.16.840.1.113883.6.96" code="195967001"><displayName value="Asthma"/></problemCode></clinicalStatement>
31
Aziz Boxwala, MD, PhDPresident, Meliorix, Inc.
Subject Matter Expert, Health eDecisions
HeD Key Deliverable 2: HL7 CDS Knowledge Artifact Implementation
Guide (IG)
(HeD Use Case 1 IG)
32
Goal
• CDS interventions must be made shareable and implementable so that they can be acquired and deployed by any organization
33
Use Case Overview
34
Design objectives
• Format for specifying computable CDS knowledge– Knowledge can be imported into existing CDS
systems• Not creating a new execution format
– Format must be flexible• Support different CDS intervention types
– More than alerts and reminders
• Knowledge must be portable
35
Approach• Reviewed several existing knowledge
representation formalisms– Did not meet requirements
completely• Solution is a harmonization of
several formats and ideas– CDSC-L3, ArdenML, CREF– HL7 Order Sets Specification,
Infobutton– GEM, HQMF, eRecommendations,
GELLO, research from SHARP-C, …
36
CDS Artifact Sharing Use Case
FR & Data Elements
VMR GEM
eRECS CDSC L3
HL7 Order
Set Model
SHARP ARDEN
Inputs
HeD Knowledge
Artifact Schema
Harmonization and ModelingHarmonization and Modeling
CREF
Knowledge artifact schema
• Modular, component based solution• Specifies key components of any CDS
intervention– Metadata– Expression language– Action– Trigger– Data model (by reference)
• vMR
37
Knowledge artifact schema
• Schemas defined by composition of components
• Currently supported CDS interventions– Event-condition-action rules– Order sets– Structured documentation templates
• In future, can be expanded– E.g., Plans of care, infobutton rules/knowledge,
relevant data display
39
Expressions• External Data
– Specifies the data required to evaluate the artifact
• Logic– Criteria used within CDS
• Calculations– Dosing, risk
40
External Data Example
<def name="PertussisProblems"> <expression xsi:type="ClinicalRequest" dataType="vmr:Problem" cardinality="Multiple“ isInitial="true" useValueSets="true" dateProperty="diagnosticEventTime.begin" codeProperty="problemCode" triggerType="DataElementAdded"> <description>Pertussis problem</description> <codes xsi:type="ValueSet" id="2.16.840.1.114222.4.11.7005" version="1"/> </expression></def>
41
Addressing the curly braces issue using HeD expression language and the VMR
Conditions Example<condition> <logic xsi:type="And"> <description>(Patient lives in SD or Care encounter was in SD) and (Diagnosed with Pertussis or Cause of Death was Pertussis or culture results positive for pertussis)</description> <operand xsi:type="Or"> <operand xsi:type="ExpressionRef" name="PatientLivesInSDCounty"/> <operand xsi:type="ExpressionRef" name="EncounterWasInSDCounty"/> </operand> <operand xsi:type="Or"> <!-- Necessary clinical conditions --> <operand xsi:type="ExpressionRef" name="HasActivePertussisProblems"/> <operand xsi:type="ExpressionRef" name="DeathWasCausedByPertussis"/> <operand xsi:type="ExpressionRef" name="HasPositivePertussisCulture"/> </operand> </logic> <conditionRole value="ApplicableScenario"/></condition>
42
Communication Action
<simpleAction xsi:type="CreateAction"> <actionSentence xsi:type="ObjectExpression" objectType="vmr:CommunicationProposal"> <property name="message"> <value xsi:type="ComplexLiteral"> <value xsi:type="dt:ED" value="This patient has or is suspected of having pertussis. Patients diagnosed with or suspected of having pertussis must be reported to County of San Diego Health and Human Services Agency within one working day of identification or suspicion"/> </value> </property> </actionSentence></simpleAction>
44
Kensaku Kawamoto, MD, PhDAssociate Chief Medical Information Officer, University of Utah
Initiative Coordinator, Health eDecisions
HeD Key Deliverable 3: HL7 Decision Support Service IG
(HeD Use Case 2 IG)
45
Goal
• Allow any organization to easily obtain CDS guidance through a secure, standard Web service interface.
46
Use Case Overview
CDS Guidance Requestor
CDS Guidance(care guidance)
CDS Guidance Supplier
CDS Request (patient data)
47
Key Standards• HL7 Decision Support Service (DSS)
– Defines SOAP and REST Web service interfaces for CDS guidance services
• HL7 Virtual Medical Record (vMR)– Provides easy-to-understand data model for CDS
• HL7 Consolidated CDA (C-CDA) and Quality Reporting Document Architecture (QRDA)– Terminology bindings and value sets largely being
adopted within vMR as vMR templates
48
CDS Guidance Service – Example
Decision Support Service
Decision Support Service EHR SystemEHR System
vMRvMR
Eval. ResultEval.
Result
49
Sample CDS Guidance ServicesEvaluation Input Evaluation Output
Medication identifier, age, gender, weight, serum creatinine level
Recommended maximum and minimum doses for medication given patient's estimated renal function
Insurance provider, data relevant to prescription
Prior authorization to prescribe medication
Patient summary Wide range of care recommendations
Patient age, gender, past health maintenance procedures
List of health maintenance procedures due or almost due
50
Sample Current Implementers• OpenCDS (www.opencds.org)
– Multi-institutional open-source effort led by Univ. of Utah– Implements HL7 DSS and vMR; will support HeD UC 2– 200+ members from 150+ organizations– Example implementation: Immunization Calculation Engine
(ICE), led by HLN Consulting, & used by New York City, Alabama, eClinicalWorks
• Enterprise Clinical Rules Service– Part of CDS Consortium effort
• Epic EHR– Will support CDS Guidance Services in 2014 release
51
OpenCDS Knowledge Editor
52
www.OpenCDS.org
53
Standards Status
54
Standard Description HL7 Status
Comments
HL7 vMR Logical Model
Base UML specification
Passed HL7 ballot Sept. 2013
HL7 vMR XML Specification
XML specification Passed HL7 ballot Sept. 2013
HL7 vMR Templates
Includes terminology binding
Passed HL7 ballot Sept. 2013Developing additional templates
HL7 CDS Knowl. Artifact IG
Use Case 1 IG Passed HL7 ballot Jan. 2013
HL7 DSS Release 2
DSS base specification
Passed HL7 ballot Sept. 2013 with 100% affirmative vote
HL7 DSS IG Use Case 2 IG Passed HL7 ballot Sept. 2013 with 100% affirmative vote
Ref: http://wiki.hl7.org/index.php?title=HL7_CDS_Standards
Victor Lee, MDVice President of Clinical Informatics, Zynx Health
Community Contributor, Health eDecisions
Pilots and Vendor Perspective
55
56
57
Do the right thing
58
59
Pilot Partnerships – HeD Use Case 1EHR Pilot Content SupplierDesign Clinicals
Order Set – Heart Failure
Zynx Health
Allscripts Rule – NQF 068 (Million Hearts)
newMentor
Allscripts Rule – San Diego Pertussis
CDC
VA Documentation Template – UTI
Wolters Kluwer Health
60
Order Set Exchange Pilot
61
WKH Doc. Template in VA CPRS
© 2013, Kensaku Kawamoto 62
63
“People who say it can’t be done
should get out of the way
of people who are doing it.”
64
Jacob Reider, MDActing National Coordinator for Health IT
Executive Sponsor, Health eDecisions
Conclusions and Panel Discussion
65
Conclusions
• True community effort• Remarkable achievements in limited timeframe• Unique opportunity to realize the promise of CDS
and make Meaningful Use of EHRs
66
Discussion Questions
• What do you think of HeD and its potential impact?• How can we best encourage CDS sharing at scale?• How can we best align CDS with healthcare
transformation?• What more can the public and private sectors do to
facilitate CDS-enabled healthcare improvement?• How can we start to make progress NOW?• What kind of content would be most useful?
67
Thank You!
Kensaku Kawamoto, MD, PhDAssociate Chief Medical Information Officer, Univ. of Utah Health Sciences CenterInitiative Coordinator, Health eDecisions
Tonya Hongsermeier, MD, MBAChief Medical Information Officer, Lahey Health
Aziz Boxwala, MD, PhDPresident, Meliorix, Inc.
Victor C. Lee, MDVice President of Clinical Informatics, Zynx Health
Jacob Reider, MDActing National Coordinator for Health IT
68