Lastic_edit.qxp 4/6/07 4:50 pm Page 15
The Convergence of Healthcare and Clinical Research Standards
Therefore, together with the FDA, CDISC and HL7 created the Regulated medical dictionary for regulatory activities (MedDRA),
9
etc. – or
Clinical Research Information Management (RCRIM) technical developed within the RCRIM Vocabulary Working Group using the
committee
3
to provide a framework for common developments. enterprise vocabulary services from the NCI.
10
The entire development
process is supported by a number of HL7 v3.0 tools to assist developers
However, early attempts to harmonise HL7 and CDISC models faced in building compliant interchange structures.
difficulties. To begin with, HL7 develops messages. By nature, a message
is transient: it contains a piece of information extracted from one place Given that CDISC and HL7 develop models and specifications, not
to be delivered to another before disappearing. This is perfectly suitable programs, putting all four pillars in place is only a signal for the software
for personal medical records that consist of various documents such as development work to begin; there are still plenty of issues to be resolved.
clinical examinations, laboratory results, X-rays, videos, etc. However, it This is common for all projects that aim to link electronic medical records
does not provide for easy access across patients or structured data with clinical research systems.
domains. CDISC, on the other hand, is built around well-defined clinical
trial protocols with strict data structures, and development is driven by Real-life Applications
the need to use technology to improve the efficiency of the heavily The interconnection of patient hospital records and clinical research
regulated clinical trial process.
4
systems faces many issues: privacy, security, non-matching patient
identification numbers, FDA regulatory compliance (e.g. 21 CFR part 11),
Of the CDISC models, ODM is designed to comply with good clinical etc. These have been reviewed and analysed several times, as can be
practice regulations – for example, audit trails and electronic signatures found in the references.
11,12
are built in – while SDTM allows easy review and analysis across studies
and across patients – for example the ability to pool all adverse events One of the most appealing solutions to these issues is the Single Source
from several drugs of the same class, or verify the exposure of a certain Project,
13
which aims to have a one-time collection of data that are
patient population to a drug across several studies. subsequently rendered into multiple formats/systems using CDISC and
HL7 standards. Single Source was tested in a CDISC proof-of-concept trial
The only CDISC model that was easily translated into an HL7 model was
the LAB model, because its primary use is the transfer of data from a
central laboratory to a clinical research database: in other words, a
Given that CDISC and HL7 develop
message. Currently, the LAB model can be implemented in four different
models and specifications, not
ways: as a CDISC XML file, an ASCII flat file, an SAS data set or an HL7
v3.0 message. programs, putting all four pillars of
interoperability in place is only a
Bringing Them All Together
The goal of information standardisation is the interoperability of
signal for the software development
computer systems, i.e. the possibility for one computer system to
work to begin…
adequately process data from another system without human
intervention. The prerequisites for interoperability are known as ‘the four
pillars of interoperability’:
5
that led to the co-development of a new methodology, or integration
profile, with the Integrating the Healthcare Enterprise (IHE) initiative.
14
1. a common reference information model; Called Retrieve Form for Data-capture (RFD), this profile enables any
2. unambiguously defined data types; electronic health record (EHR) to retrieve data-capture forms from many
3. a robust mechanism to bind domain-specific terminologies; and external systems.
4. a formal top-down message development process.
RFD was recently demonstrated at the Interoperability Showcase at the
Following this, a domain analysis model was developed for clinical Healthcare Information and Management Systems Society (HIMSS) 2007
research that is compliant with the HL7 reference information model Annual Meeting.
15
Similarly, the US National Cancer Institute (NCI) and
(RIM). This clinical model, which eventually became the biomedical Oracle collaborate on the electronic Data Collection Instrument (eDCI)
research integrated domain group (BRIDG) model with the project to develop a method of transferring the definitions and
participation of the US National Cancer Institute (NCI), supports the functionality of data capture forms from the data dictionary of a clinical
first pillar of interoperability between hospital information systems data management system into an EHR.
16
and systems used in clinical research and in regulatory agencies.
Data types used should follow or be translated into the HL7 v3.0 The Single Source approach was also recently tested at the European
specifications. Hospital Georges Pompidou in Paris with promising results.
17
Data were
shown to be imported into existing EHR and electronic data capture
Terminology is also an area that needs to be standardised. This is handled (EDC) systems using HL7 and CDISC standards, and work continues on
by two HL7 technical committees (Vocabulary and Modelling & semantic integration of both sides.
Methodology), while the vocabulary itself is either drawn from existing
domain-specific terminologies – e.g. systemised nomenclature of human Looking Ahead
medicine (SNOMED),
6
logical observation identifiers, names and codes All of these pilot projects show that progress is being made. For example,
(LOINC),
7
digital imaging and communications in medicine (DICOM),
8
HL7 v3.0 and CDISC ODM have been stable for years; the BRIDG model
DATA MANAGEMENT IN PHARMACEUTICAL RESEARCH & DEVELOPMENT 15
Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26