Integration of Health Information Resources into Electronic Health Records Using HL7 Guilherme Del Fiol, MD, MS Biomedical Informatics Department, University of Utah Intermountain Healthcare, Salt Lake City, UT James J. Cimino, MD Department of Biomedical Informatics, Columbia University, New York, NY Saverio Maviglia, MD, MSc Partners Healthcare System, Boston, MA
Outline Background HL7 infobutton standard Demonstration participants Infobutton Managers Information resource providers Live demonstration
Infobuttons Background
Information for Decision-Making ? MRSA On-line resources are much more likely to have the available information, in a bibliographic database, an electronic text book, etc
Addressing Information Needs with Infobuttons Clinical information systems evoke information needs Clinician’s computer has access to resources Context can be used to predict need Context can be used to automate retrieval
Context-Dependent Information Needs ? ! Institution Data Task Age Sex Training Role Context
Infobuttons vs. Infobutton Manager Clinical System Resource s Infobutton Manager Context Query Knowledge Base Page of Hyperlinks
Infobutton standard Overview
Why do we need a standard? There is not a common integration language Parameter names Terminologies used for content search retrieval Hundreds of resources available Not designed for infobutton integration: suboptimal results Labor intensive integration: just a few are actually used
Multiple ways of “asking” the same question What is the dose of azithromycin ? i http://resource1.com/ search = “azithromycin AND dose” http://resource2.com/ query = "azithromycin"[MeSH Terms] AND dose[All Fields] http://resource3.com/ searchConcept = 3333 ^ azithromycin filter = 11 ^ dosage
i No standard in place Resource 1 Clinical Information Infobutton API http://resource1.com/search.cgi? search = “azithromycin AND dose” http://resource2.com/ query = "azithromycin"[MeSH Terms] AND dose[All Fields] Clinical Information System Infobutton Manager Resource 2 i API API http://www.resource3.com/search.cgi? searchConcept = 3333 ^ azithromycin filter = 11 ^ dosage Resource 3 API
Standard-based integration Resource 1 Columbia HL7 HL7 Electronic Health Record Intermountain i HL7 HL7 Resource 2 HL7 Partners HL7 Resource 3 HL7
Key points XML and URL-based syntax Recommends adoption of a set of standard terminologies (e.g., RxNorm, LOINC, SNOMED-CT, MeSH) Aligned with national initiatives Flexible requirements to allow faster adoption
Example The user is looking at a problem list of a female, 94 years-old patient with Heart Failure. The user clicks on an infobutton that presents a series of questions. The user selects “How do I treat Heart Failure?”
<gender code=“F" displayName=“Female"/> <age value=“94" unit=“a"/> <taskContext code=“PROBLISTREV"/> <mainSearchCriteria code="428“ codeSystem="2.16.840.1.113883.6.103" displayName=“Heart Failure"/> <mainSearchCriteria code="428“ codeSystem="2.16.840.1.113883.6.103" displayName=“Heart Failure"/> <informationRecipient> <patient> <language code=“eng"/> <subTopic code="Q000628" codeSystem="2.16.840.1.113883.6.177" displayName="therapy"/>
<patientContext> <gender code=“F" displayName=“Female"/> <age value=“94" unit=“a"/> <taskContext code=“PROBLISTREV"/> <mainSearchCriteria code="428" codeSystem="2.16.840.1.113883.6.103" displayName=“Heart Failure"/> <subTopic code="Q000628" codeSystem="2.16.840.1.113883.6.177" displayName="therapy"/>
URL-based message Simpler implementation Support industry backwards compatibility Faster adoption Rules for automated conversion URL can be automatically derived from XML model
age.v=56 age.u=a administrativeGenderCode.c=F mainSearchCriteria.c.c=2823-3 mainSearchCriteria.c.cs=2.16.840.1.113883.6.1 mainSearchCriteria.c.dn=Serum potassium mainSearchCriteria.c.ot=K taskContext.c.c=LABRREV interpretationCode.c.c=L
http://www.e-resource.com/api? patientPerson.administrativeGenderCode.c=F age.v=56&age.u=a taskContext.c=LABRREV mainSearchCriteria.c=2823-3 mainSearchCriteria.cs=2.16.840.1.113883.6.1 mainSearchCriteria.dn=Serum potassium mainSearchCriteria.c.ot=K
Demonstration Participants
Content providers ACP PIER Clin-eguide (Wolters Kluwer Health) Dynamed (Ebsco) Lexicomp Micromedex (Thomson Healthcare) UpToDate
Infobutton Managers Intermountain Healthcare First production version in 2001 Infobutton Manager since 2005 Medication order entry, problem list, lab results 1,000+ users per month Knowledge base: resources and questions configured in XML files
Infobutton Managers Columbia University Concept of interest translated into controlled terminology Related concepts identified Topics/questions matched to concept classes and other context parameters XML table of topics (along with javascript) returned to the user Links are initiated from user’s browser
Infobutton Managers Columbia University – usage Infobuttons available since 1996 Infobutton manager version 1 2002 Available in: WebCIS: lab results, micro results, sensitivity results, inpatient drugs, outpatient drugs, problem list Eclipsys: lab orders, drug orders, nursing orders Regenstrief Medical Records System: drug orders NY State Psych Institute: drug orders NextGen: lab results 700+ users per month 2100+ uses per month
Infobutton Managers Columbia University – benefits Easy to use: 92% Question on list >50% of time: 89% Answered question: 69% Useful: 77% Helpful >50% of time: 90% Positive effect on care: 74%
Infobutton Managers Partners Healthcare Live since 2002 Medication order entry, problem list, lab results (8 clinical apps) Federated search engine for 2 library portals 50K sessions by 5K unique users per month 60% RN, 15% MD, 11% PharmD 1-50% of patient encounters 90% medication queries Median session duration under 15 seconds! 85-90% success rate Resources and context triggers configured in SQL/Access – no terminology or lexical analysis
http://www.hl7.org/v3ballot/html/ welcome/environment/index.htm