Académique Documents
Professionnel Documents
Culture Documents
The specification for the eCTD is likely to change with time. The submission should include a style sheet that supports
Factors that could affect the content of the specification include, presentation of the XML instance, navigation according to the
but are not limited to: table of contents and access to all documents within the
• Change in the content of the CTD, either through the submission. A standard style sheet is defined and provided by
amendment of information, at the same level of detail, the ICH M2 EWG. Presentation and navigation via other style
or by provision of more detailed definition of content sheets on the receiving side should be possible. The XML eCTD
and structure DTD includes attributes for descriptive names of folders and
• Change to the regional requirements for applications documents.
that are outside the scope of the CTD
2
Lifecycle Management
links from the instance to leaf files in the eCTD submission as
The applicant creates a submission that is stored in a local opposed to creating a single XML document that contains the
repository. The applicant submits the initial submission to the entire eCTD submission. The instance should contain mostly
agency, which imports the submission into another local linking facilities to the leaf files. The instance also contains meta-
repository. The nature and kind of the local repositories is not data at the leaf level. Although the specification for the eCTD is
within the scope of the eCTD. The initial submission should be very detailed, this does not mean that all eCTDs will be the same.
self-contained meaning that it includes all documents and no Again, the eCTD is a common format; the content will be different
references to other submissions. Regional guidance should be according to the application and the nature of the submission. It
consulted if references to other submissions are needed. was recommended by FDA in December, 2002 that, if a section is
Following the initial submission, the applicant can submit not populated by the submission information, the sponsor should
incremental updates such as amendments and variations. not delete that section. Instead, the sponsor should insert a brief
Updates can refer to documents in the previous submissions. explanation why it is not applicable.
Updates should be designed in a way that they can be loaded into
the repository by fully preserving the initial or previous submission
via version control. The XML backbone should include meta-data
Logical Documents and Files
identifying the update and providing navigation aids to filter for
different submission types. A logical document is comprised of one or more CTD table of
contents sections that together contain the minimum amount of
It is preferred that when a Common Technical Document is information to be exchanged. In general, the XML eCTD DTD
submitted electronically, the entire submission should be in maps explicitly to the CTD table of contents, but there are
electronic form with the exception of certain regional forms that exceptions where the XML eCTD DTD maps to the level of use
currently require written signatures. Regional requirements are designated by the appropriate ICH CTD Implementation Working
listed in appendices of the guidance. A description of how to Group (IWG) instead. Ideally, a logical document consists of a
submit a CTD containing both paper and electronic components is single physical file. In the event the physical file exceeds the
explained in appendices. recommended maximum file size due to graphics, data content,
scanned images, or other large format content, additional files
may make up the logical document. Furthermore, if the logical
eCTD Submission document consists of multiple file formats, then more than one
physical file would be needed. An example of such a case would
Generally, the eCTD Submission is a directory structure with files be PDF and XML data that together represent the logical
including the XML eCTD instance, reports, data and other document.
submission information. One major benefit is expected when the
eCTD Submission is loaded into an information system that Formats
supports the review process. One can view an eCTD Submission
from the reviewer desktop with a web browser as it is web ready.
It can be usable without processing in at least in stand-alone Formats should be readable at least for as long as it is needed for
mode with the web browser or via network connected by a server. the regulatory process. This process could be very long; e.g. 50
Another major benefit is that the entire submission has a life-cycle years. This points to neutral formats: formal standard, industrial
management structure. This structure facilitates the ability to standard, vendor independent, text-like, etc. The format should be
append, replace, or delete any module, section, or document adapted to the type of data. The list of agreed formats will be
within a section. Its flexible infrastructure complements the updated as technology evolves and new requirements arise. XML
content standards. will be the preferred format for all types of data. The common
formats that can currently be included in an eCTD Submission
are:
The eCTD Submission is composed of the following:
• Directory Structure
• XML eCTD instance • Narrative: Portable Document Format (PDF)
• Content files • Structured: Extensible Markup Language (XML)
• Graphic: Whenever possible, use PDF. When appropriate or
when PDF is not possible, use Joint Photographic Experts
Directory Structure Group (JPEG), Portable Network Graphics (PNG), Scalable
Vector Graphics (SVG) and Graphics Interchange Format
The directory structure is a structure of directories and files. There (GIF). Special formats for very high resolutions may be
should be a reasonable maximum number of entries (directories appropriate on a case-by-case basis.
and files) per directory. The directory structure should follow the • SAS datasets: SAS XPT for transported SAS datasets
rules below. The files could be in several formats as specified
below. The name of the files and directories are identifiers. They A pharmaceutical company has suggested to the FDA that SVG
should be short. The file names are not intended to convey meta- (scalable vector graphics) be used instead of PDF. PDF was
data, though some meaning in the names helps; i.e., no random approved by ESTRI in March 1997. However, it has been
names. Names for directories and files are recommended in an proposed that SVG should be used instead because 1) it is a
appendix. Any directory names and file names that are added to W3C standard 2) it is text-based while PDF is binary. The
the eCTD submission by the applicant should be descriptive and company maintains that SVG would be better for archiving and it
logical. reminds that SVG is vendor neutral. While this company makes a
strong case, no documentation could be found that this is being
considered.
XML eCTD Instance
The instance is in the submission sequence number directory. Links
The submission sequence number directory should contain at
least two files and one or more directories. One of the files in the
Links among objects in the eCTD Submission should be relative.
submission sequence directory is the instance and the other is the
The intention is to make the eCTD submission self-contained. All
MD5 checksum of the instance. The instance is the starting file for
literature references introduced by the applicant should be
the processing by an XML processor. The intention is to have
3
included in the submission, for secondary references (references Standards-Development Collaborating
to a reference), absolute links to external objects can be used.
Organizations
The capacity to point to a specific location within a file depends on
the linking technology. Different formats allow for the use of
different linking technology. • Electronic Regulatory Submission & Review (ERSR) -
a joint initiative between CDER, CBER, Office of
Regulatory Affairs (ORA), and Office of Information
Checksums and Resource Management (OIRM )
• Clinical Data Interchange Standards Consortium
The eCTD Submission should contain checksums for each (CDISC) - Open group of vendors, pharmaceutical
individual file including a checksum file for the eCTD XML companies, government agencies
instance. Initially, the MD5 Message-Digest Algorithm (MD5) • FDA subgroup of Department of Health and Human
should be used for this purpose. Including a checksum for each Services (DHHS) Data Council - Coordinates data
individual file provides a number of benefits including: standards used by FDA
• The integrity of each file can be verified by comparing • Health Level Seven (HL7) - ANSI accredited standards
the checksum submitted with the file and the computed development organization, Regulated Clinical Research
checksum. Information Management Technical Committee
• The checksum can be used to verify that the file has not (RCRIM)
been altered in the historical archive of the regulatory • Center for Disease Control (CDC)
authority. This is especially useful as the files are • National Cancer Institute (NCI)
migrated from one storage medium to another, as in the • Veterans Health Administration (VHA)
case of backup to magnetic tape storage. • DataPharm Foundation
• PharmQuest
• PPD Informatics
• Lincoln Technology
Summary – the eCTD Standard
By leveraging the use of XML and other current vendor- For the year 2003, the FDA has standards for clinical data
independent industry standards, the eCTD becomes the standard submissions among the items to be addressed. Below are the
document that is both machine-readable—so it is easily parsed highlights of the goals of the some of the above organizations as
and processed electronically—and human-readable—so it can be they work toward “normalization” of clinical data and a brief
easily retrieved and reviewed by FDA. Therefore, XML will be description of each.
easier to document, validate, audit and archive; this, in turn, will
expedite review and ease compliance. The standardization of the
eCTD is portrayed down to the directory and file levels for the Benefits of Study data standards
document. And, because of the very nature of the electronic
submission, standards are also proposed for the structure and • Improve efficiency for clinical research
components of the electronic files.
• Facilitates design and conduct of clinical trials
• Facilitates communication between researchers and
study sponsor (e.g., between CRO and drug company)
Evolving Standards • Improve efficiency of evaluation of safety and efficacy
of investigational treatments
“Normalizing” or standardizing can be related to the programming • Facilitates communication between regulatory authority
concept of macros. For example, instead of assigning variables and applicant
named BPSYS, BPDIA, and PULSE then performing the Proc • Facilitates development of efficient review environment
Univariate on each of the three variables, one can create a macro (e.g., training, analysis tools)
name of “BPRESULT”, call &BPRESULT, and pass it BPSYS,
BPDIA, and PULSE. Even though one has to allow time for Study Data Standards Initiatives
additional effort of creating the macro, the programming efficiency
would be valued as a two- to three-fold increase. Moreover, when • Study data information model/structure
this macro is re-used, subsequent programming for this algorithm
may be reduced. In general, this improves programming • File format
maintenance and ultimately improves time to market for safe and • Controlled terminology
effective treatments. Extending this idea to the standardization of • Content standards
data, file, and document structure, work is currently ongoing to
evolve standards in these areas. Extending to higher levels,
• Case report tabulations
standards initiatives are progressing in the areas of electronic
document management (eg, eCTD), study data management, and
medication information management. Standards are also Standard Descriptive Variables
developing internationally via Health Level Seven organization,
inter-regulatory via International Conference on Harmonization,
nationally via Department of Health and Human Services, and at • Observations characterized by descriptive variables
FDA via the Information Management Program. Following is a • Types of descriptive variables
discussion of the proposed standards to be included in electronic
data transfer and some of the organizations that are working
o Topic identifier - the focus of the observation
toward those standards. o Unique subject identifiers - the subject of the
observation
o Timing - describes the start and end of the
observation
o Qualifiers - describes the traits of the
observation
4
Standard for Labeling – Bar Codes • AERS gets message from Gateway and processes
• AERS-level status message is generated and sent back
• Drug and Biological Products to trading partner via the Gateway
Drug listing • Trading partner's Gateway returns receipt for AERS-
Product ingredients and packaging level status message
Patient medication identification
• Medical devices CDISC
5
Sponsors should check with their review Divisions objectives. These guidelines discuss the variety of
before deciding on whether to use the new Version 2.0 issues that should be considered when submitting
vertical format for ECG and/or Vitals or the original information to aid the statistical review of the
horizontal format. The vertical models led to some submission. One example presents guidelines for a
improvements in the respective horizontal models. familiar percent change from baseline analysis
Version 2.0 added a non-core variable for LOINC frequently used by industry statisticians.
(Logical Observation Identifiers Names and Codes)
codes for Lab, ECG, and Vitals measurements. LOINC
codes are universal identifiers for laboratory and other Laboratory Standards Team (LAB)
clinical observations (vital signs, hemodynamics,
intake/output, ECG, obstetric ultrasound, cardiac echo, The CDISC Laboratory Data Interchange Standard has been
urologic imaging, gastroendoscopic procedures, etc) developed over the past two years by the CDISC Lab Team,
which includes representatives from pharmaceutical companies,
central laboratories, contract research organizations (CROs) and
Operational Data Modeling Team (ODM) technology application developers. The Laboratory Model has
been developed by this team through multiple iterations and
tested with representative laboratory data. The Model was then
The ODM is an XML-based clinical data model designed for data
further revised to address comments provided by a 65 member
transfer and data archiving. (Details of added functionality in
Laboratory Review Committee of industry experts and a sixty day
Version 1.1 Final released November, 2001; further detail
public review. Utilization of electronic data capture (EDC)
described at PharmaSUG 2002 and at CDISC website.)
technologies and standards is a major issue across the industry.
The CDISC LAB Model Version 1.0 Final was released for
• Ability to address changes to key data values implementation and comment in November 2002 and replaces an
• Expanded transaction support for partial or incremental earlier public comment version that was released in June 2002.
transfers and archives
• Expanded metadata descriptions of more complex
event structures
• Support for including multiple studies and reusable FDA subgroup of HHS Data Council
metadata in one file
• Support for depicting non-patient reference data
The purpose of the bar code regulation for labeling is to reduce
• Support for vendor extensibility the number of adverse drug events, including deaths, which occur
• Increased compatibility with the CDISC Submissions every year due to medication errors. It is hoped that bar code
Data Model labeling for human drug products can minimize errors due to drug
• Increase support for archiving of clinical data and mix-ups and also provide crucial information for prescribers. Bar-
metadata coding may ultimately apply to biologicals, medical devices, and
• Elimination of support for the flat representation of animal drugs. Electronic submissions of labeling will allow
clinical data included in version 1.0 computer matching of labeling products and product-labeling
• Corrections to numerous bugs including those related changes that will greatly enhance the accuracy and speed of
to time-zones, locales and signatures review.
• Changes to the order of elements
• Changes to element names and attributes, including the
item data element Health Level 7 (HL7)
In December, 2002, CDISC submitted to FDA the following. We “Level Seven" refers to the highest level of the International
believe the FDA Guidance should directly reference use of the Standards Organization's (ISO) communications model for Open
CDISC Operational Data Model (ODM) as a standard format for Systems Interconnection (OSI) - the application level. The
archiving operational clinical trial data at the investigator’s and/or application level addresses definition of the data to be exchanged,
sponsor's site for possible review by FDA auditors and other the timing of the interchange, and the communication of certain
interested parties. ODM is a vendor neutral, platform independent errors to the application. The seventh level supports such
open standard that is supported by CDISC and offered to all functions as security checks, participant identification, availability
interested parties without cost. ODM recreates the original data checks, exchange mechanism negotiations and, most importantly,
collection environment experienced by the investigator or clinical data exchange structuring. While HL7 focuses on addressing
data management operator, describes all CRFs with its questions immediate needs, the group continues to dedicate its efforts to
and range checks, and produces a complete audit trail of all ensuring concurrence with other United States and International
changes to the data. ODM supports the representation of standards development activities. A special interest group has
electronic signatures with data, and provides for inclusion of an e- been appointed to assist in the implementation of the
CRF form similar to those commonly included with eSubmissions Administrative Simplification provisions of HIPAA mandates,
based on data collected by Electronic Data Capture (EDC) providing on-going support, and to represent HL7 in the HIPAA
systems. ODM requires no proprietary software tools, which may Designated Standards Maintenance Organization (DSMO) efforts.
become obsolete over time, so it is an ideal format for retaining One of its constituent committees is the Working Group for Data
electronic clinical data at investigator sites. CDISC is also Interchange (WEDI). The purpose of the DSMO is to encourage
working closely with HL7 to harmonize the ODM with HL7 the use of HL7 for uniform implementation of this supplemental
standards. We hope the FDA will consider the applicability of the information. This group coordinates industry input to produce and
CDISC ODM XML standard as a platform and vendor maintain guides for HL7 electronic messaging. Another HL7
independent solution for archive and maintenance of clinical trial initiative is intended to be the basic unit of a document-oriented
data. Electronic Patient Record (EPR). The Clinical Document
Architecture (CDA), which was until recently known as the Patient
Analysis Dataset Model (ADaM) Record Architecture (PRA), provides an exchange model for
clinical documents (such as discharge summaries and progress
notes)—and brings the healthcare industry closer to the
This document provides guidelines for the creation of realization of an electronic medical record. It is the first XML-
analysis data sets and associated documentation that based standard for healthcare. Using XML ensures that PRA
are submitted to the FDA statistical reviewer in support documents are readable using: a) widely-available and commonly
of the primary and important secondary study deployed XML-aware browsers and b) a generic PRA style sheet
6
written in a standard style sheet language. The Clinical Document
Architecture was approved as an ANSI standard November 2000. S ta n d a r d s
Although the CDA has been approved, it has, and is creating
much controversy over its potential for breech of patient
C o n verg en ce
information confidentiality.
www.fda.gov/ohrms/
www.phrma.org/
7
ACKNOWLEDGMENTS
I would like to thank Matt Becker and Monte Jarvis for their
support.
CONTACT INFORMATION
Susan Bairnsfather
PharmaNet, Inc.
Cary, NC 27513
(609) 951-6583
sbairnsfather@pharmanet.com
SAS and all other SAS Institute Inc. product or service names are
registered trademarks or trademarks of SAS Institute Inc. in the
USA and other countries. ® indicates USA registration.