Grace Paterson, Medical Informatics Xiaoli Wang, Dalhousie Computer Science

Electronic Exchange of Structured Interim Discharge Summaries Using the XML-

based Clinical Document Architecture

Grace Paterson, Medical Informatics

Xiaoli Wang, Dalhousie Computer Science

http://www.medicine.dal.ca/dmedinfo

XML Standard for Healthcare

Comprehensive healthcare information exchange standard must include the full electronic health record, not just fielded data

Clinical Document Architecture (CDA) is a specification for exchanging clinical documents using eXtensible Markup Language (XML)

Bob Dolin’s Ehealth 2001 Workshop presentation available on CIHI HL7 website

Power of XML

XML, like HTML, is a derivative of Standard Generalized Markup Language (SGML)

Information Technology’s “lingua franca” Current “must-have” feature in software

Browsers, Notepad, SAS, Oracle, Internet Security E-Commerce and Automating Transactions

E-Health gets a tremendous boost at low cost

Power of a Think Tank - Kona Proposal

Week of July 7, 1997, a group of physicians, healthcare system

vendors and consultants met at Kona Mansion, NH

OUTCOME: a multi-level architecture for the exchange of

Electronic Health Care Records

April 2000. Kona Proposal was revised into the PRA (Patient

Record Architecture) but went to HL7 ballot August 2000 as

CDA (Clinical Document Architecture)

Xiaoli phoned Liora Alschuler for specifications

CDA and Shared Care

CDA gives priority to documents generated by clinicians involved in direct patient care

Care occurs where the patient’s pillow is - hospital, family physician, long term care

CDA standard can be readily implemented, and is platform and application independent

Promotes shared care in appropriate setting

Structured Discharge Summary

QEII DEPARTMENT OF MEDICINE Header Information (Participants and Roles) Most Responsible Diagnosis/Admitting Dx Comorbidities/Cardiac Risk Factors Allergies Course in Hospital Pertinent Investigations/LabResults

Structured Discharge (con’d)

Follow Up Recommendations for Family Doctor Medications on Discharge (unchanged from

admission, altered, new) Discharge Outcome Measures Physician’s Signature/Status/Print Name Discharge Summary and Dictated Job #

CDA and Knowledge Integration

Publishers submit abstracts to MedLine in XML XML tag <FullText> provides external link to

fulltext for MedLine abstract CDA has <content>, <link>, <coded_entry>,

<observation_media>, and <local_markup> Nova Scotians can link from CDA document to

Electronic Bookshelf entries in doctorsns.com for prompts and “Information Given to Patient”

CDA and Outcomes Research

• From time of Florence Nightingale to mid-60s, Hospital Annual Reports documented discharges as recovered, improved, no change, worsened, died

• Generic scales (SF-36) • Disease specific scales (IBDQ)• Health Status (comfort, function, lifespan) • Health Outcomes and HL7 V3 Data Types

FeedbackModel

Key features of CDA A clinical document should be human readable. The clinical document will be legally authenticated, and the

authentication of a clinical document will be applied to the whole, not portions of the document.

Entrusted person or organization will maintain the clinical document.

A CDA document can include text, images, sounds and other multimedia content.

CDA documents are encoded in XML. CDA documents will be consistent with HL7 data types.

Necessary and Sufficient Components

Persistence

Stewardship

Potential for Authentication

Wholeness

Human Readability

SpecificationOverview of CDA Architecture In Level One, the level one DTD is for all kinds of clinical

document. In Level Two, a specific clinical document will be defined in

consultation with Professional Societies QEII Hospital has gone through up to 12 revisions of Cardiology,

Internal Medicine, Geriatrics, COPD templates We were able to implement each template in Level One

Defined process for going from CDA Level One to CDA Level Three - tightly coupled with HL7 V3 Data Types

Deliberate decision to wait for V3 ballot results Aug 2001

XML Design (Header) CDA Header is used to uniquely identify a clinical document in order to

exchange it among organizations There are four components for header: document information, Encounter data,

providers and patients Document information includes <id>, <set id>, <version_nbr>,

<document_type_cd>, <confidentiality_cd>, <document_relationship> Encounter data describe the setting in which the documented encounter

occurred. It includes <patient_encounter>, <practice _setting_cd>, <encounter_tmr>, <service_location>, <addr>.

Provider include the person who participated in the services being documented.

Patient include the patient and other significant participants (such as family members)

XML (Level One Body)

Nested containers in Level One body: sections, paragraphs, list and tables.

Minimal amount of markup and minimal constraint for this markup.

Coded entries uses HL7 Version 3 Data Types Appendix A lists object identifiers, e.g.,

2.16.840.1.113883.6.3 is the external coding scheme ICD10

<coded_entry><section>

<caption>Most Responsible Diagnosis</caption>

<section>

<caption>Unstable Angina

<caption_cd V=“I20.0” S=“2.16.840.1.113883.6.3”/>

</caption>

<paragraph>

<content>Y</content>

</paragraph>

</section>

Object Identifier for ICD10 coding scheme

ICD10

Implementation To generate a Discharge Summary, the user

enters the information into Web form Information is POSTed to Web server where a

Java servlet parses the input data and marks up the data into the CDA format

Written to disk as an XML file Retrieved as XML file by Java servlet and parsed

into constituent elements by SAX parser, converted to HTML and displayed

Implementation

XML describes the meaning of content, independent of its display

There are two style sheets for XML CSS (Cascading Style Sheet) XSL (XML Style Sheets Language)

User enters chart number for search and choice of display style (web form or discharge summary)

Implementation

Parser of XML We must access information in XML

documents through an XML parser. There are two kinds of parser: SAX and

DOM, SAX is event-based parser and DOM is object-based parser, the core object is Node.

Implementation

• Several events for SAX:• startDocument()/endDocument()

• startElement()endElement()

• Characters()

• Several methods for DOM:• NodeType, parentNode, childNode, firstchild,

getElementByTagName(), getText(), getNameItem().

Summary and Next Steps The procedure of this project includes understanding

specification, designing a system, implementing the system, and the final testing.

The purpose of the system is to meet the need of an extensible, hierarchical, structured clinical document exchange.

Work-in-progress with other CS Students: CDA document storage in Statistical Analysis System (SAS) CDA integrated with Electronic Bookshelf Automatic Indexing Using UMLS for Coded Entries

Acknowledgements• Dr. Michael Shepherd, Computer Science

• Dr. Carolyn Watters, Computer Science

• Dr. David Zitner, Director, Medical Informatics; Chair, Health Records Committee, QEII Health Sciences Centre

• Kathy MacNeil, Director, Patient Information Services, QEII Health Sciences Centre

• Patient Care Record Committee, Capital Health District Authority

• Mary Eileen Wall, Clinical Informatics Coordinator, QEII Health Sciences Centre

• Sandra Cascadden, Director of IT Services, QEII Health Sciences Centre

• Ron Soper, Computer Science CO-OP Student