1 © T O U C H B R I E F I N G S 2 0 1 2
Clinical Trials
Improving Data Quality in the Clinical Trial System
Julia Zhang, PhD
Julia Zhang, PhD, is Associate Director, Global Biomedical Informatics at Genzyme, a Sanofi Company. Dr Zhang has worked in the
pharmaceutical and biotechnology industries for many years. Dr Zhang is an active member of the standard organizations Health Level
Seven International (HL7) and Clinical Data Interchange Standards Consortium (CDISC). Her expertise is in following areas:
program/project management; regulatory submissions both in paper and electronic form; development/implementation of industry
standards (HL7, CDISC) for clinical trial and submission; clinical trial data management, including clinical database design, electronic data
capture (EDC), data reviewing, data cleaning, and data exchange; and clinical trial data processing, statistical analysis, and reporting. Her
broad clinical experiences across different therapeutic areas include infectious diseases, respiratory diseases, diabetes, oncology, and
rare inherited disorders.
The pharmaceutical industry is facing unprecedented challenges. For
example, the blockbuster effect has been highlighted by patent
expirations, poor outcomes, globalization, aging, and decreasing
healthcare funding. Recent research summarizes that, to produce just
one Food and Drug Administration (FDA)-approved drug, it
takes around 10–15 years, costs about $1 billion, and involves
1,000–6,000 people.1–3 To face all the current challenges, the real need
is to improve innovation and pharmaceutical R&D productivity.
Organizations make strategic and operational decisions based on
data. As John Martin, CEO of Gilead, said in an interview with Fortune
in May 2011: “We tend to think of the pill as the drug. But the drug is
really the data that’s been collected through clinical research that
defines the best use of that drug.”
Poor-quality data can negatively influence how a company is perceived
in the marketplace. Therefore, it is critical to ensure data quality is given
the highest priority. The drug development process can be treated as an
international trip: data quality is the passport for reaching the final
destination—an approved drug. Improving enterprise process efficiency
and data quality is critical for the enterprise’s return on investment (ROI).
Data Quality
Data quality is reflected by multiple factors, such as the method used and
time taken for data collection, and the formats and types of data collected.
The following principles should apply to high-quality data:
• accuracy—the degree to which data correctly reflect the real world
object or verifiable sources. All data values should be within the value
domains specified by the business;
• consistency—data should be synchronized across the enterprise. For
example, data should be consistent between systems and no duplicate
records should exist;
• integrity—data should have a complete or whole structure. All
the characteristics of the data, including business rules, data
relationship rules, dates, definitions, and lineage, must be correct for
data to be complete;
• timeliness—data should be available at the time needed; data
delayed is data denied;
• completeness—it is possible that the data are not yet available but
still considered completed because they meet user expectations; and
• auditability—this means that any transaction, modification, and report
can be traced back to the originating transaction.
Data quality is not linear; having data quality on one dimension is as good
as ‘no quality’. Poor-quality data increase costs through wasted resources
(due to rework or lack of processes), generate a need to correct and deal
with reported errors (rework), and lead to an inability to optimize business
processes. Poor-quality data will result in lost revenue through customer
Disclosure: The author has no conflicts of interest to declare.
Correspondence: julia.zhang@genzyme.com Citation: Touchhealthsciences.com; January, 2012
Acknowledgment: Special thanks to Sue Dubman and the Genzyme GetSMART program teams.
Poor-quality data can negatively influence how a company is perceived in the marketplace. Therefore it is critical
to ensure data quality is given the highest priority. Improving data quality in a clinical trial system involves many
processes and factors. This article discusses how to improve data quality by implementing general standards. The
discussion includes how to develop business governance; data design (keeping the enterprise vision in mind);
setting a strategy for data collection, processing, and reporting; and implementing standards to develop a data
validation process. Using a well-designed data validation tool will improve data quality and ease communications.
Zhang_v1_iHC US 04/01/2012 12:21 Page 1

Improving Data Quality in the Clinical Trial System
2T O U C H H E A L T H S C I E N C E S . C O M
dissatisfaction, reduced employee morale, and poorer decision-making.
Data quality represents the state of completeness, validity, consistency,
timeliness, and accuracy that makes data appropriate for a specific use.4
Improving data quality consists of improving the processes and
technologies that are involved in ensuring the conformance of data values
to business requirements and acceptance criteria.
Clinical Trial Systems
Clinical trial systems support clinical trial development in a number of
ways, including protocol creation, case report form (CRF) design, clinical
data collection, processing and analysis, and regulatory submission. To
ensure high-quality data, clinical trial systems should be able to:
• force an organization to think about its long-term data and information
needs, and how they relate to its long-term success;
• motivate actions in the right direction—i.e. toward quality;
• provide a sound basis for decision-making within and outside
the organization;
• maximize use of the organization’s data and information, avoid
duplication, facilitate partnerships, and improve equity of access; and
• maximize integration and interoperability.
A clinical trial system includes several tools and functions, such as
a protocol development tool, a data collection and verification tool
(e.g. electronic data capture [EDC]), a statistical computational
environment for data processing, analysis and reporting, and
a regulatory submission function. The clinical trial system runs through
the clinical trial data life-cycle and consists of many complicated
processes and contributions from different sources. Communicating
effectively between the systems, tools, and data is achieved through
interoperability by implementing standards.
Improving Data Quality
To improve data quality in a complicated clinical trial system, we
first need to identify how the data were designed, collected, and
processed, and then gain a thorough understanding of the clinical
data life-cycle. In general we can start with the design and then
move on to the development and implementation. To efficiently
operate a clinical trial system that includes many different
subsystems, tools, and functions, interoperability has an important
role. Interoperability, which is fundamental to business, is the ability
to communicate between different information technology systems
and software applications, to exchange data accurately, effectively,
and consistently, and to use the information that has been
exchanged. Interoperability is increasingly important, as ‘the
network is the computer’ (a phrase reputedly coined by John
Gage and Bill Joy of Sun Microsystems in 1984) becomes a reality.
Some examples of what interoperability can do in a clinical trial
system include:
• facilitating the understanding, characteristics, and management of
data usage over the data life-cycle;
• evaluating data relationships automatically, allowing multiple
systems to exchange information without a human being having to
tell a computer how the data items are related;
• allowing multiple systems to move and share data transparently
between functions within organizations, suppliers, external partners,
and regulatory bodies;
• supporting integrated information analysis, resulting in better design
of targeted therapies;
• providing improved efficiency in the planning and conduct of clinical
trials; and
• enhancing coordination between clinical trial sites, sponsors, and
regulatory agencies.
Without interoperability, information would remain in proprietary silos.
Interoperability and information exchange are best understood as
business concepts rather than technical concepts. Interoperability
represents a major challenge because of the difficulty of integrating data
from different sources. The current clinical trial system landscape
contains a multitude of information systems, lacks a common data
model, and has few common data formats and vocabularies.
Furthermore, there is no infrastructure for data sharing. These hurdles
make interoperability difficult. To move and share data transparently
between functions within organizations, suppliers, external partners, and
regulatory bodies, interoperability becomes extremely important.
Interoperability starts with standards, but it is much broader than
standards alone. Using standards can increase process efficiency and
effectiveness, improving compliance and saving clinical trial data
process life-cycle resources (time and money). Clinical Data
Interchange Standards Consortium (CDISC) standards (see Table 1)
have been developed to support the streamlining of processes
within medical research, from the production of clinical research
protocols through to reporting and/or regulatory submission, as well
as warehouse populations and/or archives and post-marketing
studies/safety surveillance.
Table 1: CDISC Standards Updates
Standards Implementation/Status
SDTM SDTM version 1.2; SDTM IG version 3.1.2
ODM Version 1.3.1
Define.xml Version 1.0
TDM Under development
Lab Base model version 1.0.1; final XML schema version 1.0.1;
Microbiology Extension & Reference Range Model Review Version
ADaM ADaM version 2.1 and ADaM IG version 1.0;
ADaM validation checks; ADAE is for public reviewing
Protocol Version 1.0 review
Controlled Package 6 is for public reviewing
Terminology
CDASH Version 1.1
SEND SEND version 3 and SEND IG version 3.0
SHARE
Therapeutic • Alzheimer’s disease/mild cognitive impairment data standard
Area Standards • Standard cardiovascular endpoint definitions
• Parkinson’s disease
BRIDG Version 3.0.3
ADaM = Analysis Data Model; ADAE = Analysis Dataset for Adverse Events;
BRIDG = Biomedical Research Integrated Domain Group; CDASH = Clinical Data Acquisition
Standards Harmonization; CDISC = Clinical Data Interchange Standards Consortium;
IG = implementation guide; ODM = Operational Data Model; SDTM = Study Data
Tabulation Model; SEND = Standard for Exchange of Non-clinical Data; SHARE = Shared
Health and Clinical Research Electronic Library; TDM = Trial Design Model.
Zhang_v1_iHC US 04/01/2012 12:21 Page 2