Networking and Health Information Exchange: Supporting Standards for EHR

Application

Audio Transcript

Slide 1
Welcome to Networking and Health Information Exchange, Supporting Standards
for EHR Application. This is Lecture b.

This component, Networking and Health Information Exchange, addresses what is

required to accomplish networking across and among disparate organizations who have
heterogeneous systems.

Unit 7 covers Supporting Standards for EHR Application and consists of four lectures.
Over these four lectures, we will talk about the additional standards that are available to
support interoperability across different applications that relate to or are interactive with
the Electronic Health Record.

In lecture b, we will look at several guideline standards, but will focus on the HL7
Guideline Interchange Format (GLIF) standard and the ASTM Guideline Elements
Model (GEM) standard. We also will have a brief introduction to GELLO an object-
oriented expression language for clinical decision support. The discussion on GLIF is
included only for illustrative purposes. It is in very limited use, and it is being replaced by
guidelines written in GELLO.

Slide 2
The Objectives for this unit, Supporting Standards for EHR Application, are to:

Understand the clinical decision support standard Arden Syntax,

Understand standards for clinical guidelines,
Understand object-oriented expression language for clinical decision support
GELLO,
Understand the clinical decision support standard Infobutton,
Understand disease management, and
Understand other clinical decision support applications.

Health IT Workforce Curriculum Networking and Health Information Exchange 1

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
Slide 3
Additional Objectives for this unit, Supporting Standards for EHR Application, are to:

Understand other standards that help to support networking and reporting

requirements as well as functionality to optimize the connectivity among
heterogeneous systems deployed within a single enterprise,
Understand single sign-on standards and the HL7 Clinical Context Object
Workgroup (CCOW) standard,
Understand regulatory standards, and
Understand issues relating to person identifiers, master patient indices, and
record locator services.

Slide 4
Clinical practice guidelines serve to reduce inappropriate variations in practice, to
improve patient safety, to improve health care, and to reduce costs. Although the
importance of guidelines is well-accepted, the use of guidelines is still limited, and the
conformance in use (typically less than 50%.). Guidelines are typically integrated into
the workflow and involve direct, real-time interaction. Guidelines may also provide
automated reminders and alerts. Guidelines support order entry, and may be an asset
to the consistent capture of data.

We are interested in standards for guidelines in order to share computer-interpretable

guidelines.

Guidelines deliver patient-specific recommendations. They may be integrated with

EHRs to provide, among other things, automated reminders and alerts, to do decision
support and task management, to do order entry appropriateness, referral criteria, and
to do background monitoring, care plans, and quality review.

Slide 5
Benefits of guidelines include the ability to provide automatic decision support that can
be applied to individual patients, and to perform retrospective analysis to verify that
patients were treated appropriately.

Guidelines guide simulations and aid human visualization through interactive, dynamic
displays of guideline pathways. Guidelines allow one to focus on relevant sections of a
flowchart.

Guidelines may be applied to individual patients, integrated clinical data, and knowledge
to guide the encounter. Physicians are often biased by recent experiences guidelines
support consistency. Guidelines also reveal the logic behind decision making and can

Health IT Workforce Curriculum Networking and Health Information Exchange 2

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
provide a visualization of that logic. Guidelines capture the experience of experts and
bring equality to patient care.

Slide 6
Sharing guidelines invokes many challenges:

Local adaptation of guidelines including

Availability of resources and expertise,
Local workflow issues, and
Practice preferences.

Integration with local information systems by

Matching patient data from EHR to GL, and by
Matching recommendations in guideline to actions in order-entry systems.

Dissemination formats.

Perhaps the biggest barrier to the use of guidelines is the concept of cook-book
medicine. Guidelines only guide the provider in what is common practice. Many
guidelines are incomplete and dont fit the individual patient who the provider is seeing.
Most guidelines are adapted to the local environment because of the preference of the
providers, local cultures, and the nature of the patients seen. Guidelines should be
interfaced with the EHR to use existing data.

Guideline representation is critical for both computer implementation as well as the

ability for sharing clinical guidelines.

Slide 7
The potential for the use of guidelines is huge. Guidelines are part of meaningful use.
If guidelines are produced by domain experts, the length of time from research to
routine use can be significantly shortened. Guidelines must be shared in a technology-
neutral format that they can run on different platforms and systems. Guidelines should
be implemented using the service-oriented architecture: Each guideline becomes an
encapsulated, enclosed component with a defined set of input parameters and
delivering a defined set of outputs.

The development of guidelines is also enhanced by tools to support authoring,

validation, and maintenance.

Health IT Workforce Curriculum Networking and Health Information Exchange 3

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
Slide 8
Guidelines need to be expressive to cover all the nuances of patient care and decision
making. For computer interpretation, the guidelines must be structured and include
definitions, data requirements, recommendations and actions, and logical, decision-
making algorithms.

Guidelines are based on an expressive decision model. Inputs and outputs must be
clear in specifying the work to be accomplished. Actions must be expressive, complete,
and clear. Data must be interpreted correctly. Guidelines are also used to generate
alerts and reminders.

Slide 9
Guidelines must show the underlying logic when queried. This explanation is most
often expressed as a block diagram, or flow diagram, showing data requirements and
decision points.

Guidelines must be able to handle the complexity of a clinical event, and must support
the variations in patients and settings. Guidelines must support various actions. When
a provider is not compliant with a guideline, the system should request the reason. For
example, a foot exam for a diabetic patient may not be conducted because of a request
from the patient. That fact should be noted so a follow-up action can be initiated.

Guidelines must include the authoritative source for the decision logic a literature
reference, the guideline expert source, or other authority.

Slide 10
The introduction of guidelines into the work flow is often awkward. For example, if a
patient qualified for the use of three guidelines, each guideline would be presented to
the provider consecutively. If the same piece of data were required in each guideline, it
would be asked three times. If data is available from the EHR, it should be
automatically added and not require the provider to reenter the data. An even better
approach would be to take the guideline requirements and merge them into the order-
entry approach.

One challenge in the sharing of guidelines is the matching of data elements identified in
the shared guideline and the local setting.

Health IT Workforce Curriculum Networking and Health Information Exchange 4

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
Slide 11
One workable approach for the use of guidelines in mapping guideline data
requirements to the local setting is through the use of a virtual medical record (vMR). In
this case, data is extracted from the local system and placed in a virtual record in the
data position defined by the guideline. The guideline can now function interoperably
with any system. This process divorces the data requirements and the logic from the
local environment. The output can go through a similar procedure, where the output
from the guideline is mapped into the local terminology. There is, however, the risk of
the loss of information. HL7 has a group developing standards for this approach (vMR).

The success of guideline use depends on how effectively the use of the guideline is
integrated into the workflow. If the guideline is perceived to take additional time, or
require a disruptive action on the part of the provider, the guideline is likely to be
unused.

Slide 12
The National Guideline Clearinghouse (NGC), sponsored by AHRQ, makes available a
number of guidelines to providers at no cost. As of February 20, 2012, the NGC
contained 2319 individual guidelines. Guidelines are indexed by a number of
parameters including category, disease, creating group, and others. This web site is one
of the best resources for obtaining clinical guidelines.

Slide 13
This slide shows a list of common guideline representation models. We will discuss the
first three in detail (Arden Syntax, GLIF and GEM).

DILEMMA represents guidelines as a set of protocols within which actions are encoded.
EON uses activities, actions and provides advice as a patient scenario and discusses
activity states.
PROforma uses an enquiry action for information collection.
Asbru uses a similar approach to DILEMMA.
GUIDE has a wait action as its knowledge representation is in the form of Petri nets.
PRODIGY uses a similar approach to EON.

EON and GLIF use decision steps; PROforma, GUIDE and PRODOGY use decisions;
Arden uses logic slots; DILEMMA uses state transition; and Asbru uses condition-
preference as the decision-making process.

However, none of these guideline executable models have come into widespread use.

Health IT Workforce Curriculum Networking and Health Information Exchange 5

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
Slide 14
One of the most popular representation models for guidelines is GLIF, developed by Dr.
Bob Greenes and colleagues. GLIF is a format for sharing clinical guidelines
independent of platforms and systems. It is based on an object-oriented logical model
of concepts, and uses XML syntax. It is an executable model. GLIF serves only as a
model today. Further development has been discontinued.

Attempts to promote GLIF as a guideline executable standard failed because of the

inability to get agreement among several models. The HL7 Clinical Decision Support
Work Group decided to focus on GL modeling elements, such as the expression
language (GELLO); the data model which is the origin of the virtual medical record
(vMR) project; and possibly a workflow description formalism (which never generated a
project), that all GL models could use, rather than the GL model itself.

There was also a lot of time spent on narrative GL markup models like GEM, but none
of these were expected to lead to GL execution models. Also, many of the CDS WG
came to recognize that it would be very rare indeed that a GL execution model would
ever be used in practice, except for strict clinical protocol situations - but rather that a
GL would be executed in terms of components like single-step rules, order sets, etc.
That is, it would be decomposed rather than run as is.

Because of all these factors, interest in an executable GL model waned. It seems

unlikely that this activity will be sustained. This model is included in the discussion,
however, to illustrate the kinds of things these standards must address.

Slide 15
The GLIF model, similar to Arden Syntax, identifies the title of the guideline and the
author. Also similar to Arden Syntax, the model includes decision steps, action steps,
branch steps, synchronization steps, and a patient sleep step.

Slide 16
By now, what these terms mean should be obvious.

An action step might be to order a lab test or to prescribe a medication.

A decision step is a conditional or decision point in which the flow branches depending
on the result of a condition. If the patient has pain, then action.

Branch steps permit changing flow depending on decision logic. Synchronization steps
permit arresting action until a step is complete; for example the entry of the result of a
lab test and patient-state sleep provides for a continuation of the guidelines over a

Health IT Workforce Curriculum Networking and Health Information Exchange 6

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
period of time, including multiple encounters. Patient-state sleep characterizes the
patients clinical state.

Slide 17
GLIF provides three representation states.

The 1st is from the perspective of the author or viewer of the guideline: the conceptual
flow of decisions and actions. This view meets the condition of explaining the logic of
the guideline, and aids in human understanding.

The 2nd is the abstract machine representation: The computer executable form in which
correctness can be analyzed.

The 3rd level is the integration of the guideline into the workflow environment.

Slide 18
The abstract machine representation is similar to that of the Arden Syntax. The logical
expressions and actions are based on defined steps based on medical ontology. The
guideline defines values, constraints and time lines. The guideline can be interpreted
and analyzed for correctness. Tools provide syntax checking, type and range checking
of data elements.

Slide 19
The Guideline Elements Model, or GEM, was developed at Yale University, Center for
Medical Informatics, by Dr. Rick Shiffman. It is now an ASTM standard and has
continued to be developed. It uses a hierarchical data structure to organize the
heterogeneous information contained in practice guidelines and uses an XML editor
designed specifically for guideline markup. The XML editor is cleverly named the GEM
cutter.

Slide 20
This slide shows the hierarchical structure of GEM.

Logically, it uses defined data elements, conditional logic, decision variables, and logical
constructs with resulting actions. The overall structure defines data terms and decision
algorithms. Details of GEM are available from ASTM including data element definitions
used in the guidelines (available as a pdf).

Health IT Workforce Curriculum Networking and Health Information Exchange 7

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
GEM is defined as a Document Type Definition (DTD). A Document Type Definition
defines the legal building blocks of an XML document. It defines the document structure,
with a list of legal elements and attributes.

A DTD can be declared inline inside an XML document, or as an external reference.

Slide 21
GELLO is a class-based, object-oriented (OO) language that is built on existing
standards. GELLO expression language is based on the Object Constraint Language
(OCL), developed by the Object Management Group. Relevant components of OCL
have been selected and integrated into GELLO to provide a suitable framework for
manipulation of clinical data for decision support in health care.

The GELLO language can be used to:

Build up expressions to extract and manipulate data from medical records.
Construct decision criteria by building up expressions to reason about
particular data features or values. These criteria can be used in decision-
support knowledge bases such as those designed to provide alerts and
reminders, guidelines, or other decision rules.
Create expressions, formulae, etc. for other applications.

Use cases for GELLO include medical experts, researchers, home economists,
epidemiologists, knowledge engineers, and programmers. It is the language that
permits the application of knowledge to data of use in multiple clinical applications,
including clinical guidelines, template constraints, alerts, and recommendations.

Slide 22
The syntax of the GELLO language can be used with any object-oriented data model. In
the context of clinical decision support, such an OO data model can be any Refined
Message Information Model (R-MIM) view of the HL7 RIM. An example of an R-MIM
view of the HL7 RIM is the "virtual medical record. The vMR functions as a limited view
of the multiple classes in the HL7 RIM, showing only those classes relevant to the
clinical decision support application. A major problem to sharing clinical knowledge is
lack of a common format for data encoding and manipulation.

Based on the premise that GELLO can fully provide expression support for any properly
defined view of the HL7 RIM, the development of GELLO is independent of any
particular specification of an OO data model. It is thus only necessary, when producing
a set of decision support applications using GELLO, to specify the particular object-
oriented model used. Thus, GELLO addresses many of the problems of the lack of
semantic interoperability as a result of data structures. GELLO can accommodate any
properly-defined OO data model.

Health IT Workforce Curriculum Networking and Health Information Exchange 8

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
Slide 23
GELLO products are targeted to clinicians who need to use expression language for
sharing and manipulating knowledge in medical context

GELLO is becoming increasingly popular. The reasons include:

GELLO is:
A declarative language,
Extensible,
Vendor independent,
Platform independent,
Object-oriented and compatible with vMR,
Easy to read and write,
Side-effect free, and
Flexible.

GELLO is easy to use, read and write.

Slide 24
This slide shows the relationship of GELLO to other key components in decision
support. GELLO serves as a vehicle to tie together the different areas including Arden
Syntax and GLIF (representing guidelines and sequential knowledge); RIM
(representing the information model and coupling ontology); and other data storages
and knowledge databases.

The advantage of GELLO is that it can accommodate existing work including the Arden
Syntax and GLIF that represent sequential knowledge.

GELLO uses the HL7 RIM as a link to interoperability, and, finally, GELLO can provide a
link to many of the other knowledge representation models mentioned earlier in this
lecture.

Slide 25
GELLO expressions are text strings satisfying GELLO language specifications. The
result is an expression that can be evaluated for actions. GELLO accommodates
temporal actions.

GELLO uses the basic data types: Boolean, real, integer and string. It supports a wide
range of mathematical operators, including trig functions, logical comparator operators,

Health IT Workforce Curriculum Networking and Health Information Exchange 9

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.
math operations, and string operators. It provides considerable power in interacting
with data elements.

GELLO syntax is a context-free grammar consisting of a number of productions. Each

production is formed by two parts: the left-hand side consisting of a non-terminal symbol
and a right-hand side formed by a sequence of one or more non-terminal and terminal
symbols. GELLO uses a Backus-Naur Form (BNF) syntax.

GELLO Examples of expressions are:

calcium.notEmpty() and phosphate.notEmpty()
Renal_failure and calcium_phosphate_product > threshold_for_osteodystrophy

GELLO also includes temporal operators.

Slide 26
This example looks for the presence of Azotemia Observation within the last three
months.

Assumptions:
1. The data model has as code a generic term such as SNOMED "finding"
("246188002") and the value slot has the code for azotemia.
2. For a diagnosis such as azotemia, the effective time is the time interval during which
the disease is thought to be present.
3. A PointInTime.NOW() function returns the current time.

The specific GELLO representation for this logic is shown on this slide.

Slide 27
This concludes Lecture b of Supporting Standards for EHR Application.

This lecture has discussed guideline representation models. The future of such work is
likely to be based on GELLO. Guidelines themselves are very important. Use of
guidelines should increase significantly with pressure from meaningful use. IOM
released a study on Clinical Guidelines in 2011.

Slide 28
No audio.

End.

Health IT Workforce Curriculum Networking and Health Information Exchange 10

Version 3.0/Spring 2012 Supporting Standards for EHR Application
Lecture b
This material Comp9_Unit7b was developed by Duke University, funded by the Department of Health and Human Services, Office
of the National Coordinator for Health Information Technology under Award Number IU24OC000024.