ComputerBased Patient Records and Medical Informatics Standards

1
Computer-Based Patient Records and Medical
Informatics Standards
Medical Decision Support Systems

• Yuval Shahar M.D., Ph.D.

2
The Medical (Patient) Record

• A historical record of patient care
• A communication tool among care providers
• A research and knowledge-gaining tool
• A teaching tool
• An operational tool (e.g., order entry)
• A business tool (e.g. to support billing)
• An administration record (e.g., to manage
  resources)
• A legal record with considerable longevity

3
Electronic Medical Record (EMR)

• AKA Computer-Based Patient Record (CPR)
• Provides multiple advantages vs. manual records
• Record can be used by multiple personnel at the
  same time
• Record is accessible from anywhere (even from
  home)
• Clear, well-organized, legible documentation
• Data can be reused for other purposes
• Data can be integrated from multiple sources
  transparently
• Data can be validated automatically
• Enables multiple automated research and
  decision-support functions (analysis, machine
  learning and data mining, automated diagnosis,
  reminders, guideline-based care)
• Decision support can be integrated with use of
  the patient record

4
EMR Costs

• Large initial set-up investments
• Hardware, software, training, support,
  maintenance
• Significant workflow changes
• Significant organizational changes
• Difficult data entry relative to handwriting
• Potential catastrophic failure
• Note paper records also have down times

5
Integration of EMR and Decision Support Modules

• Decision support is most effective when
  integrated with an EMR
• The most likely opportunity for providing
  decision support is when the physician is
  assessing the patient record or entering an order
• All or most relevant patient data can be
  accessible to the DSS and do not require separate
  entry
• Physician should always be able to override the
  recommendation and, if relevant, provide feedback

6
Order Entry

• A major function of an EMR system, allowing care
  providers to enter clear, legible orders for
  patient care anytime, anywhere
• Supports validation of order, issuing of alerts,
  suggestion of relevant information and knowledge,
  and even actions
• Quick effect on physician ordering behavior

7
EMR and Knowledge Sources

• The most effective time to provide access to
  knowledge is when the care provider is browsing
  the patient record
• A query can be formulated in a context-sensitive
  manner with respect to the patient record, thus
  anticipating the physicians needs
• Note Queries often have relatively expected
  structure and content (e.g., which drug is useful
  for condition X in context Y What are side
  effects of drug Z when used in manner W What
  clinical guidelines are most relevant for disease
  D in patients of type P)

8
EMRs Major Issues

• Data Entry
• Data capture the scope of the data that is or
  can be represented in the EMR
• Data input coded data are difficult to input by
  physicians text is less useful for processing
• Errors can be reduced by multiple validity checks

9
Validity Checks During Data Entry in an EMR

• Range checks (Hemoglobin in 0..30 Gr/Dl)
• Pattern checks (a telephone number pattern)
• Numeric and other inter-data constraint checks
  (total of WBC differential is 100)
• Consistency checks (pregnant male??)
• Temporal-abstraction checks (weight cannot change
  by 50 Kgs in 2 days)
• Spelling checks

10
Physician-Entered Data

• The main challenge to EMR developers!
• Patient histories, physical findings,
  interpretations, diagnostic and treatment plans
• Several very different entry methods
• Transcription of dictated or written notes
• Structured encounter forms from which notes are
  transcribed and even encoded
• Direct entry of data by physician via computer
• Speech recognition might alleviate some of the
  difficulties

11
Security Issues in EMRs

• Authorization
• Is my dentist allowed to see my gynecological
  record?
• Which fields of my record can my or another GP
  view?
• Who has asked to view my records last month?
• Authentication
• Is this user really my physician?
• Encription
• Can an eavesdropper understand the message sent
  to my doctor?
• Eventually, security depends on people

12
The Need for Standards

• EMRs and almost any other information-oriented
  system in a clinical environment cannot be used
  without well-defined standards for representing
  and communicating information
• Data need to be exchanged between multiple,
  heterogeneous systems and might be used by very
  different applications
• Standards are needed for several different uses
• Identifying patients, providers, health-care
  plans, employers
• Transferring patient data across different
  systems
• Representing medical knowledge that can be
  reused

13
How are Standards Developed?

• Ad hoc
• A group of interested people and organizations
  agree on an informal specification (ACR/NEMA
  DICOM)
• De facto
• A single vendor creates standard through monopoly
  (Microsoft Windows)
• Government mandate
• Agency creates a standard and legislates it (HCFA
  UB92 claim form)
• Consensus
• A group of volunteers work openly to create
  standard (HL7).

14
Examples of Information-Standards Organizations

• American National Standards Institute (ANSI)
• Private, nonprofit
• Accredits organizations that create standards
• Technical Committee 251 (CEN TC 251)
• The European Standardization Committees
  technical committee for medical informatics
  standards
• American Society for Testing and Materials (ASTM)
• Largest non-government source of standards in the
  USA
• ASTM committee E31 is responsible for development
  of medical information standards

15
Terminologies and Controlled Vocabularies

• Precoordinated
• All concepts encoded beforehand no possibility
  of creating new terms
• Postcoordinated
• New combinations can be formed from existing
  terms to describe new concepts

16
International Classification of Diseases (ICD)

• Intended mostly for talking about dead people
  (reporting mortality statistics to the WHO)
• Strict hierarchy with core 3-digit codes,
  possibly 4th digit
• ICD-9 (1977) common inadequate for clinical
  reporting
• ICD-9-CM (Clinical Modifications) adds extra
  levels of details by 4th and 5th digits,
  popular in USA
• ICD-10 (1992) exists, but no clinical
  modifications yet

17
Codes in The International Classification of
Diseases (ICD-9 CM)
724 Unspecified disorders of the
back 724.0 Spinal stenosis, other than
cervical 724.00 Spinal stenosis, unspecified
region 724.01 Spinal stenosis, thoracic
region 724.02 Spinal stenosis, lumbar
region 724.09 Spinal stenosis,
other 724.1 Pain in thoracic spine 724.2 Lumbago
724.3 Sciatica 724.4 Thoracic or lumbosacral
neuritis 724.5 Backache, unspecified 724.6 Disor
ders of sacrum 724.7 Disorders of
coccyx 724.70 Unspecified disorder of
coccyx 724.71 Hypermobility of
coccyx 724.71 Coccygodynia 724.8 Other
symptoms referable to back 724.9 Other
unspecified back disorders
18
Diagnosis-Related Groups (DRGs)

• A USA (Yale) abstraction of the ICD-9-CM codes
• A small number of codes grouping multiple
  diagnosis codes by similar expected costs of
  hospitalization
• Modifies the major diagnosis by associated
  conditions, severity, and procedures to determine
  specific DRG code

19
Current Procedual Terminology (CPT)

• Encodes diagnostic and therapeutic procedures
• Adopted in the USA for billing and reimbursement
• Similar to DRG, classifies procedures by cost and
  reasons
• CPT-4 The main code used for reporting physician
  services to government and private insurance
  reimbursement

20
Diagnostic Statistical Manual of Mental Disorders
(DSM)

• Published by the American Psychiatric Association
• Provides nomenclature as well as definitions
  (diagnostic criteria) of psychiatric disorders
• Coordinated with ICD e.g., DSM-IV is coordinated
  with ICD-10

21
Systemized Nomenclature of Medicine (SNOMED)

• Developed by the American College of Pathologists
• Evolved from SNOP, A multi-axial system for
  describing pathological findings by
  postcoordination of topographic (anatomic),
  morphologic, etiologic, and functional terms
• SNOMED III 11 axes, more than 130,000 terms
• SNOMED-RT (Reference terminology) created to
  encourage more consistent use of terms
• Main problem Too expressiveseveral ways of
  defining the same term (e.g. acute appendicitis)

22
Read Clinical Codes

• Developed by James Read during the 1980s
• Adopted by the British National Health Service
  (NHS) in 1990
• Version 3 is a multiple hierarchy, and version
  3.1 added ability for postcoordination of
  modifiers
• Work undergoing to map to SNOMED

23
The Unified Medical Language System (UMLS)

• A project of the National Library of Medicine
  (within the National Health Institutes NIH)
• Main resource The Metathesaurus
• contains over 330,000 terms
• relates terms from over 40 different sources
• Supports searching the medical literature
• Uses Medical Subject Headings (MeSH) which are
  used to index medical literature

24
Logical Observations, Identifiers, Names and
Codes (LOINC)

• A naming system developed by McDonald and Huff
  for tests and observations (now includes also
  vital signs, ECG, etc)
• Uses six semantic axes to encode the test, such
  as substance measured (urine) and analysis method
  used
• Coordinated development with the European
  Clinical Data Exchange Standard (EUCLIDES)
  standard

25
Data Interchange Standards

• Allow a sender to transmit data (a transaction
  set) to a receiver in unambiguous fashion
• Closely related to the Open Systems
  Interconnection (OSI) 7-layer communication model
  of the International Standards Organization (OSI)
• Physical, data link, network, transport, session,
  presentation, and application (semantic-specificat
  ion) layers
• Typically use position dependent or tagged field
  format

26
Example Data-Interchange Standards

• ACR/NEMA
• American College of Radiologists with the
  National Electronic Manfacturers Association
• Current version DICOM 3.0 uses an object
  oriented model and supports ISO communications
• ASTM E31
• Published E1238, Standard Specification for
  Transferring Clinical Observations Between
  Independent Systems
• E1460 Defining and Sharing Modular Health
  Knowledge Bases is the Arden Syntax for Medical
  Logical Modules

27
Health Level 7 (HL7)

• Today, includes more than 500 industrial and
  academic organizational members and over 1800
  individual members
• Name refers to OSI application layer 7
• A standard for exchange of data among different
  hospital computer applications
• Built upon ASTM 1238 and other protocols
• Version 3 (1999) is object oriented and uses a
  Reference Information Model (RIM)

28
Functions of a Health-Care Information System
(HCIS) (I)

• Patient management
• Admission, Discharge, Transfer (ADT)
• Patient tracking
• Departmental management
• Ancillary departmental systems support clinical
  departments laboratory, radiology, pharmacy,
  blood bank and medical records are most commonly
  automated
• Care delivery and Clinical documentation
• Mostly order entry and results reporting

29
Functions of a Health-Care Information System
(HCIS) (II)

• Clinical decision support
• Built upon other HCIS components and need to be
  integrated with them (e.g. during order entry)
• Financial and resource management
• Typically the first functions to be centralized
• Managed-care support
• Integrated Delivery Networks (IDNs) start
  focusing more on patient health maintenance
  rather than cutting costs of treating sick
  patients
• Thus, provider-profiling systems, contract
  management systems and more sophisticated modules

30
Three Classic HCISs (1)

• The HELP system at the University of Utah
• Developed by Warner et al. at LDS Hospital
• Incorporated decision support logic modules from
  the start these react to data and issue
  reminders, alerts, and advices
• Uses the HELP Frame Language
• Eventually led to Medical Logical Modules and the
  Arden Syntax

31
Three Classic HCISs (2)

• The Center for Clinical Computing (CCC) system at
  Beth Israel Deaconess Medical Center
• Developed by Bleich and Slack as a centralized
  system in Beth Israel Hospital, Boston from 1978
• Intensively used
• Includes knowledge access to MedLine via the
  PaperChase module, as well as email
• Ambulatory system supports problem lists and
  clinic notes
• Uses a MUMPS database, used as the clinical-data
  repository, and the ClinQuery online data
  warehouse
• Very little decision-support functionality

32
Three Classic HCISs (3)

• The DIOGENE System at Geneva Canton University
  Hospital
• Developed by Jean-Raoul Scherer and colleagues
  from 1971
• Migrated from a centralized to distributed
  architecture
• Supports all administrative and clinical
  functions
• Reports are printed physicians write orders by
  telephoning an operator who types the order while
  physician dictates, views typing on computer
  screen, and gives verbal consent.