HM312 Week 6

© 2017 American Health Information Management Association© 2017 American Health Information Management Association
Health IT and EHRs: Principles and
Practice, Sixth Edition
Chapter 10: Data Infrastructure
Assessment

© 2017 American Health Information Management Association
Data Infrastructure
• Data infrastructure refers to what data are
needed to operate an enterprise and how
they are defined (vocabulary), structured and
processed (architecture) and quality-
assured.
• Data architecture for health IT supports
ability to create the data-information-
knowledge-wisdom continuum.

Data Architecture supports D-I-K-W
• Data = raw facts and figures that make up communication
• Information = data that have been combined to produce value
• Knowledge = information enhanced with experience
• Wisdom = knowledge with insight
• Heuristic thought is processing of data by humans that gives
them their intelligence
• Knowledge management is a discipline associated with those
who work primarily with their minds (knowledge workers).
• Learning organization is one where knowledge management is
central to organizational performance.

Types of EHR Data

Formats of Data Stored in Computer
• Structured data
o Values of variables
o Stored in databases
o Can have significant operations performed on them
• Reflections of original data (aka image data,
unstructured data)
o Narrative text
o Video & audio
o Images
o Computers enhance their availability and access

Both Types of Data are Necessary
• Structured data support clinical decision
making; but narrative data support clinicians’
understanding of the patient story
• Data entry aids are helping to blend structured
and unstructured data.
• Natural language processing (NLP) would
convert narrative data to structured data.
o Improving in maturity, but due to contextual nature
of healthcare data still difficult to fully achieve

Discrete Reportable Transcription
• Combines dictation of narrative notes with NLP
that tags data elements so they can be placed into
structured data collection templates
Traditional
Dictation
Note
Produced
Speech
Dictation
Structured
Data for
EHR
Note
Transcribed
(or Reviewed)
Follows EHR
Template
System Tags
Data for
EHR

Vocabulary Standards
• Codes - Representation of words to enable
machine processing
• Classification, or taxonomy - Grouping of terms
with similar meanings used for a specific purpose
• Vocabulary, Terminology, and Nomenclature
o Vocabulary – all terms within a domain
o Terminology – prescribed set of terms
o Nomenclature – system of naming
• Language - System of communication
• Data mapping - process of identifying
relationships between two distinct data models, which
may be used to coordinate data among different
classification systems, mediate between sources and
destinations of data, and when transitioning from one
version of a system to another
• Vocabulary server - software that enables
multiple vocabularies to be used across different
applications
Broad
Specific

Data Mapping Goal and Examples
Ultimate goal:
Capture clinically specific data
Once at the point of care, and
Derive information there from for
Every other legitimate use
Primary Purpose Secondary Use Mapping From Mapping To
Clinical documentation Service reimbursement SNOMED CT ICD-10
Lab orders Billing LOINC CPT
Documentation of
ADE/ADR
Regulatory reporting SNOMED CT MedDRA
Clinical problem list Literature search for
decision support
SNOMED CT MeSH

Codes and Coding
• Codes are used to represent words in machine processing
o Codes may be structured into a classification system (e.g., ICD-10-
CM), or be random representations of words or concepts (e.g.,
SNOMED CT)
• Coding is the process of assigning codes to words.
o Medical coding (with ICD-10-CM and CPT) has largely been a
manual process. Note: automated code books that help a coder
locate codes is not computer-assisted coding.
o Computer-assisted coding using NLP can assign codes from an
EHR. This is the primary way in which SNOMED CT codes are
assigned.
o Note also that ‘coding’ can refer to the development of software,
where code refers to representation of the instructions in a computer

Code Sets and Data Sets
• Code set refers to a group of associated codes.
o Most medical classifications (e.g., ICD-10-CM, SNOMED CT, CPT)
are code sets.
o Where an entire language or terminology has a set of codes, the set
of codes is generally called a lexicon.
o Code sets exist for many types of data used in healthcare; and not
all are medical code sets. For example, there is a Claim Adjustment
Reason Code (CARC) set that is used to describe why changes
have been made in reimbursement from what is requested on a
claim. Another common code set is the Zip Code set.
• Data set is a predefined list of data that need to be
collected for a registry or special data set. The data
collected may or may not be encoded.

Certified EHR Technology Code Set Requirements
• ICD-10-CM or SNOMED CT are the code sets required for
problem lists
• LOINC is a code set required for documenting lab data
o May be used for other observation data such as vital signs and nursing
data
• RxNorm is a group of code sets required for describing
medications, developed by the National Library of Medicine,
Veterans Administration, and Food and Drug Administration.
o Vendors providing these code sets (and often accompanying clinical
decision support for drug alerting) include: Multum, Micromedex, First
Databank, Gold Standard Drug Database, and MediSpan
o Also included in RxNorm is the VA’s terminology (National Drug File-
Reference Terminology [NDF-RT]) used to code clinical drug properties

SNOMED CT
• SNOMED CT is a clinical reference terminology
o Enables consistent capture of detailed clinical information
o It is largely used to code concepts, descriptions, and relationships
o Originally developed by the College of American Pathologists as a
multi-axial system to describe the etiology, topography,
morphology, and function of pathological tissue; later adding other
axes to form Systematized Nomenclature of Medicine (SNOMED)
• Today, SNOMED CT is an international standard maintained by The
International Health Terminology Standards Development
Organization, based in Denmark
• College of American Pathologists provides SNOMED Terminology
Solutions that aid:
o Implementing SNOMED CT into systems
o Building SNOMED CT subsets
o Extending content (guidance on extensions)
o Modeling content
o Mapping local code sets to SNOMED CT

SNOMED CT Concepts
• SNOMED CT has over 344,000 concepts with unique
meanings and definitions organized into hierarchies.
• A description table contains more than 913,000 English-
language and 660,000 Spanish language descriptions or
synonyms for flexibility in expressing clinical concepts.
• A relationship table contains approximately 1.3 million
relationships to enable reliability and consistency of data
retrieval.

Example of a SNOMED CT Code
• 284196006: Burn of skin
o 246112005 (Severity) = 24484000: Severe
• 113185004: Structure of skin between fourth and
fifth toes
• 272741003 (Laterality) = 7771000: Left

Other Classifications & Terminologies
• ABC Coding Solutions for complementary medicine
• International Classification of Functioning, Disability, and Health
• MEDCIN is a proprietary vocabulary primarily for physician office use to
describe symptoms, history, physical exam results, and other data
• MedDRA is a Medical Dictionary for Regulatory Activities
• Nursing Terminologies (see next slide)
• National Drug Code (NDC) is a universal product identifier for drugs
• Unique Device Identification (UDI) helps encode information in
medical device adverse event reporting
• Universal Medical Device Nomenclature System (UMDNS) is an
international standardized nomenclature and coding system relating to
unique medical device concepts and definitions

Nursing Terminologies
• American Nurses Association recognizes nursing
terminology and supports their mapping in SNOMED
CT.

Unified Medical Language System (UMLS)
• National Library of Medicine (NLM) provides the nation’s principal
biomedical bibliographic citation database, MEDLINE/PubMed.
• To index its journals for the database, it developed the Medical
Subject Headings (MeSH) controlled-vocabulary thesaurus.
• NLM has been a strong supporter of facilitating the development of
EHRs, distinguishing between:
o Semantics - the study of meaning, including ways meaning changes over
time
o Syntax - the study of patterns of formation of sentences and phrases from
words and grammar
o For effective use of EHRs, the meaning of terms and their format must
work together

UMLS Knowledge Sources
• Aid retrieval and integration of biomedical
information from bibliographic databases, EHRs,
and other sources
• These include:
– UMLS Metathesaurus links over 100 biomedical
vocabularies and classifications
– SPECIALIST Lexicon contains syntactic information for
terms not in the Metathesaurus
– UMLS Semantic Network contains information about
concepts and their permissible relationships

Data Architecture
• Specific way each individual data element is used in
the information system
o Data sets
• Predefined group of data elements
o Data registries and data registry functionality
• Registries:
o Separate databases existing apart from a provider’s EHR, and often
outside of a given provider setting
o Examples: cancer registries, immunization registries
• Registry Functionality - functions that can be performed on a panel of
patients simultaneously, rather than one-by-one. Registry functionality in an
EHR enables the EHR to process data from a registry
o Big data refers to the massive amount of data available to study

Standardized Data Sets
• Uniform Hospital Discharge Data Set (UHDDS)
• National Quality Forum (NQF) measures
• ORYX (Joint Commission)
• Healthcare Effectiveness Data and Information Set (HEDIS)
• Continuity of Care Record (CCR) from ASTM International
• Many others

Databases
• Databases:
A data structure for
information
processing
Files of related
information
• Database
Management
Systems (DBMS)
Software and data
structure to support
databases
Types of
databases
• Flat file
• Hierarchical
• Relational
• Object-oriented
• Multi-dimensional
• Hadoop
• Data repository
o Relational database designed with an open
structure not dedicated to software of any
one vendor, which collects and organizes
data to provide an integrated,
multidisciplinary view
o Used for online transaction processing
(OLTP)
o May also be called:
• Transactional database
• Operational database
• Data warehouse
o Hierarchical or multi-dimensional database
that collects data on which complex
analysis is performed
o Used for online analytical processing
(OLAP)
o May also be structured into data marts and
operational data stores

Data Repository
• Primary means to collect and provide data for
transactions performed in an EHR
• Requires the following data integration functions:
o Data transformation
o Data cleansing
o Linkage
Copyright © 2012, MargretA Consulting, LLC. Reprinted with permission.

Data Warehouse
• Collection of data that can be reorganized into
more suitable formats for ad hoc querying and
analytical processing
• Data warehouse management system (DWMS)
extracts data from a repository or application
database and applies data integrity routines to the
data so they are suitable for the type of processing
to performed in the warehouse:
o Data normalization eliminates redundancy
o Data denormalization creates intentional redundancies to support
multiple uses, often in segments of the data warehouse (i.e., data
marts)

Data Warehousing

Data Management
• Data Modeling
o Entity-relationship
o Relational
o Object
• Data Dictionary
o Captures the results of data
modeling
o Supplies metadata (data
about data)
• Knowledge Representation
o Processing data to support
clinical decision making
o Ontology is the
representation of knowledge
in a given domain
•Metadata (ISO/IEC 11179
standard
o Descriptive metadata
• Describes data elements to be captured
and processed in an application
• Describes data attributes
• Provides processing rules
• Identifies relationships among data
• Provides keys (or links) to a data model
• A database (called a data dictionary)
usually is used to store this metadata
(see next slide)
o Structural metadata
• Describes how the data for each data
element are captured, processed,
stored, and displayed. A data model is
used for this purpose (see following
slides)
o Administrative metadata
• Metadata programmed into the software
to be generated by the software.
• Provides information about how and
when data were created and used.
• Examples:
o Audit log of access to data
o Decision support rules used to alert EHR
users of potential issues with a patient
o Data provenance identified where data
have originated from and where data may
have moved between databases

Data Dictionary and Example Entry

Data Model Examples

Knowledge Representation
• Encoding of knowledge on computers to enable
systems to reason automatically (“machine
learning”). Examples:
o Artificial intelligence (such as Amazon suggests other
products based on your past buying patterns)
o Expert systems (such as clinical protocols developed
with data from a very large number of patients)
• Ontology is a structural framework, or
representation of knowledge, that helps model and
create knowledge

Data, Information, and Knowledge
Governance
• Governance is the establishment of
policies and continual monitoring of their
proper implementation for managing
organization assets to enhance the
viability of the organization
o Key assets include data
• Governance processes ensures quality
data and data collection strategies

HM312 Week 6

More Related Content

What's hot (10)

Similar to HM312 Week 6 (20)

More from BealCollegeOnline (20)

Recently uploaded (20)

HM312 Week 6