Home-Grown Coding Systems A Critical Step in EMR

Home-Grown Coding Systems —A Critical Step in EMR Implementation Eric Rose, MD Associate Director for Clinical Informatics Information Systems Department University of Washington Physicians Network http: //faculty. washington. edu/momus/ December 10, 2002 Eric Rose, MD 1

SUMMARY n n n December 10, 2002 Implementation of any EMR (including vendor-supplied) requires the that the user institution create coding systems This can be done well or badly It matters Eric Rose, MD 2

Some definitions n n December 10, 2002 Concept: an idea encompassing a class of objects ("unit of knowledge created by a unique combination of characteristics"-ISO) Term: A word denoting a concept Terminology/Controlled Vocabulary/Ontology: a set of terms pertaining to a given domain, not necessarily with any structure Nomenclature: A terminology “structured systematically according to pre-established naming rules” (ISO) Eric Rose, MD 3

Some definitions (cont’d) n n December 10, 2002 Coding system: A terminology + context-free symbolic codes for each term Classification/taxonomy: A terminology system + specified relationships between terms (“concept system”-ISO 1087 -1) Eric Rose, MD 4

What is a coding system and what does it do? n n n December 10, 2002 Represents in a standardized fashion Groups Separates Abbreviates Facilitates automated dataprocessing & transmittal Eric Rose, MD 5

Types of coding systems n n December 10, 2002 Simple, 1 -1 (CA, NY, TX) Categorical (record store bins) Hierarchical (homo sapiens) Multiaxial (Dewey-decimal, SNOMED) Eric Rose, MD 6

What does this have to do with EMR Implementation? December 10, 2002 Eric Rose, MD 7

THE UNIVERSE OF CONCEPTS READ LOINC SNOMED CT CPT NANDA ICD-10 NDC ICD-9 -CM NCPDP December 10, 2002 Eric Rose, MD 8

What does this have to do with EMR Implementation? n n December 10, 2002 Most EMR’s allow customized choices for various database items Each one of these is a small coding system Eric Rose, MD 9

What does this have to do with EMR Implementation? (cont’d) If developed carefully, home-grown coding systems facilitate: Intuitive data entry u Interpretable data @ individual patient level u Usable data at population level u Usable data for automated decision-support systems u Data that is shareable with other systems u December 10, 2002 Eric Rose, MD 10

Examples of mini-coding systems you might need to create n n n n December 10, 2002 Disease Categories for Family History Reason for Visit Allergic Reaction Type Delivery Outcome Anesthesia Type for Surgery Source of Diagnostic Specimens Ethnic Group Marital Status Eric Rose, MD 11

What makes a coding system good? n n n Completeness Nonredundancy Clarity Stability Granularity appropriate to use or flexible Evolutionary (Adapted from Cimino, 1998; Chute et al. , 1998) December 10, 2002 Eric Rose, MD 12

Completeness There should be a term for any possible situation December 10, 2002 Eric Rose, MD 13

Completeness—Example Reason for Medication Discontinuation Allergic response Alternate therapy Availability Cost of medication Discontinued by another Health Care Provider Discontinued by patient Dose adjustment Duplicate Error Ineffective NON Covered Medication Paradoxical response Pregnancy Prescription never filled Reorder Resistant Organism Side effects Therapy completed What if medication was never taken by patient? No way to denote that! December 10, 2002 Eric Rose, MD 14

Nonredundancy There should be only one term for any given situation December 10, 2002 Eric Rose, MD 15

Nonredundancy-Example Next of Kin-Relationship to patient Domestic Partner Life Partner Significant Other December 10, 2002 Eric Rose, MD 16

Clarity The categories in your coding system should be unambiguous to all users December 10, 2002 Eric Rose, MD 17

Clarity-Examples Family Medical History category “Blood Disease, ” “Anesthesia” Medication reason-for-d/c “Alternative Therapy, ” “Error” December 10, 2002 Eric Rose, MD 18

Stability Once defined, the meaning of a code must not be changed, though it may be inactivated so it is not applied to any new cases December 10, 2002 Eric Rose, MD 19

Appropriate or Flexible Granularity = Level of detail described by the coding system, i. e. “fineness” of categorization Low granularity = Few, broad categories High granularity = Many, narrow categories December 10, 2002 Eric Rose, MD 20

Appropriate Granularity. Examples Family History categories “Alcohol dependency, ” “Drug dependency”—It is sufficient to just have one category for “Chemical Dependency” “Heart disease”-Not granular enough to meet needs of risk assessment for coronary disease December 10, 2002 Eric Rose, MD 21

Evolutionary Coding system development is an ongoing process, requiring addition of new categories and inactivation of old ones to keep the system congruent with prevailing ideas. Example = Family History category, “Venous Thrombosis” December 10, 2002 Eric Rose, MD 22

Guiding principles n n n December 10, 2002 CENTRALIZE control of the coding system Keep your lists SHORT Respond PROMPTLY to user requests for additions and explain rationale when it’s not appropriate to meet the request Design for the future including new user types & interfaces Careful with “Other” Eric Rose, MD 23

For further reading: n n December 10, 2002 Bakken S et al. Toward Vocabulary Domain Specifications for Health Level 7 -coded Data Elements. JAMIA 7: 333 -342, 2000. Cimino JJ. Desiderata for controlled medical vocabularies in the twenty-first century. Methods Inf Med. 1998 Nov; 37(4 -5): 394 -403 Chute CG. Cohn SP. Campbell JR. A framework for comprehensive health terminology systems in the United States. JAMIA 5(6): 503 -10, 1998 Nov-Dec. ISO 1087 -1: 2000. Terminology Work. Vocabulary-Part 1: Theory and Application Eric Rose, MD 24