Clinical trials are conducted to answer scientific questions.
Does a
treatment work?
Is it safe?
How does it
compare with existing treatment?
Which
patients benefit?
These
questions can only be answered reliably when the underlying clinical data are
trustworthy.
A trial may
enrol hundreds of participants, follow a sophisticated protocol and use
advanced statistical methods.
But if
critical data are inaccurate, incomplete, inconsistent or cannot be
reconstructed, the scientific conclusions may be compromised.
This is why
data integrity is fundamental to clinical research.
It is not
simply a responsibility of the data management department.
It is a
shared obligation across investigators, sponsors, CROs, technology providers,
laboratories, medical monitors, statisticians and quality teams.
ICH E6(R3)
reinforces the importance of data governance throughout the trial lifecycle,
including appropriate controls over data collection, processing, review,
transfer, retention and computerized systems [1].
What Is
Data Integrity in Clinical Trials?
Data
integrity refers to the completeness, consistency, accuracy and reliability of
data throughout their lifecycle.
In
practical terms, it means being able to establish:
What was
recorded?
Who
recorded it?
When was it
recorded?
Where did
it originate?
Was it
changed?
Why was it
changed?
Can the
information be reconstructed?
These
questions are central to confidence in clinical trial evidence.
1.
Understanding ALCOA+
The ALCOA
principles provide a useful framework for understanding data integrity.
The
original ALCOA elements are:
|
Principle |
Meaning in clinical research |
|
Attributable |
The person or system responsible for the record can be identified |
|
Legible |
The information can be read and understood |
|
Contemporaneous |
Information is recorded at the appropriate time |
|
Original |
The original record or an appropriately verified copy is preserved |
|
Accurate |
The record correctly represents the observation |
The
commonly used ALCOA+ extension adds:
- Complete
- Consistent
- Enduring
- Available
Together,
these principles provide a practical framework for evaluating whether clinical
data can be trusted.
They are
not a substitute for applicable regulations or study-specific controls, but
they help teams understand the characteristics of reliable records.
2. Data
Integrity Begins at the Source
Consider a
participant whose blood pressure is measured during a study visit.
The
observation may initially be recorded in:
- A hospital electronic health record
- A clinical worksheet
- A device-generated record
- A dedicated electronic source system
It may then
be transferred into an electronic Case Report Form.
The
reliability of the final dataset depends on the integrity of the original
observation and the processes used to transfer or transform it.
If the
original measurement is inaccurate, subsequent verification cannot make it
scientifically correct.
This
illustrates a fundamental principle:
Data
quality cannot be repaired solely at the end of a study.
It must be
supported at the point where information is generated.
3. Source
Data and EDC Are Not Always the Same
Electronic
Data Capture (EDC) systems are widely used to collect trial data.
But the EDC
system is not necessarily the original source for every data element.
For
example:
Source
record: Hospital laboratory system
EDC record: Laboratory
result entered into the study database
Alternatively,
data may be entered directly into an electronic system that constitutes the
source record.
Sponsors
should clearly define source data locations and authorized data originators.
FDA's
guidance on electronic source data specifically addresses traceability,
authorized source data originators, data capture and investigator
responsibilities [2].
4. Audit
Trails Protect Traceability
An audit
trail provides a record of relevant actions affecting electronic data.
For
example:
A
participant's laboratory value is initially entered as:
5.2 mmol/L
It is
subsequently changed to:
4.2 mmol/L
A suitable
audit trail should enable authorized reviewers to determine, as applicable:
- The original value
- The revised value
- Who made the change
- When it was made
- Why it was made
Without
traceability, it may be difficult to determine whether the change corrected an
error or introduced one.
ICH E6(R3)
emphasizes appropriate audit-trail functionality, retention and review within
computerized systems [1].
5.
Electronic Systems Must Be Fit for Purpose
Clinical
trials increasingly rely on interconnected systems, including:
EDC
CTMS
eTMF
IRT/RTSM
eCOA/ePRO
Laboratory
systems
Imaging
platforms
Safety
databases
Electronic
health records
These
systems may generate, modify, transfer or retain important clinical trial
information.
Sponsors
need assurance that systems are suitable for their intended use and that
relevant risks are controlled.
Depending
on the system and applicable requirements, controls may include:
- Validation or documented fitness-for-purpose
assessment
- User-access management
- Role-based permissions
- Audit trails
- Data backup and recovery
- Change control
- Security
- Data transfer verification
- Business continuity
- Record retention
FDA's
October 2024 final guidance on electronic systems, records and signatures
provides recommendations for ensuring electronic records and signatures are
trustworthy and reliable in clinical investigations [3].
6. Data
Transfers Create Additional Risk
Modern
trials often transfer data between systems.
For
example:
Central
Laboratory → Data Management
eCOA → EDC
Imaging
Vendor → Clinical Database
EDC →
Statistical Programming
Each
transfer creates opportunities for:
- Missing records
- Incorrect mappings
- Duplicate data
- Incorrect units
- Misaligned participant identifiers
- Date or time errors
- Version inconsistencies
Data
transfer specifications and appropriate verification procedures are therefore
important.
The
objective is to ensure that information remains accurate and traceable as it
moves between systems.
7. Missing
Data Can Threaten Scientific Conclusions
Not all
data-quality problems involve incorrect values.
Sometimes
the most important issue is data that were never collected.
Consider a
clinical trial in which participants with worsening symptoms are more likely to
discontinue follow-up.
If outcome
data are disproportionately missing from these participants, the remaining
dataset may not adequately represent the intended study population.
This can
introduce bias.
Missing-data
prevention therefore begins with:
- Appropriate protocol design
- Realistic visit schedules
- Participant retention strategies
- Clear data collection procedures
- Early identification of missing assessments
- Appropriate statistical planning
Data
integrity is therefore connected to both operational execution and statistical
validity.
8. Data
Cleaning Is Not the Same as Data Integrity
Data
cleaning involves identifying and resolving discrepancies, inconsistencies and
incomplete information.
Examples
include:
- Implausible values
- Missing dates
- Inconsistent visit information
- Incorrect units
- Duplicate records
- Conflicting clinical entries
These
activities are necessary.
But data
cleaning cannot necessarily correct:
- Fabricated source records
- Unreliable endpoint assessments
- Systematic measurement errors
- Unrecorded clinical events
- Poorly defined source-data processes
Data
integrity requires prevention and governance in addition to cleaning.
9. Clinical
Data and Safety Data Must Be Reconciled
A clinical
trial may capture adverse events in EDC while serious adverse event reports are
processed in a separate safety database.
These
systems may contain related information.
Discrepancies
can arise in:
- Event terms
- Onset dates
- Seriousness criteria
- Outcomes
- Participant identifiers
- Follow-up information
Appropriate
reconciliation helps ensure that clinically important discrepancies are
identified and resolved.
This
requires collaboration between data management, pharmacovigilance, medical
monitoring and clinical operations.
10. Data
Integrity Is a Cross-Functional Responsibility
Different
teams contribute to different stages of the data lifecycle.
Investigators
and site staff generate and document clinical observations.
CRAs review
relevant site processes and critical data according to the monitoring plan.
Data
managers manage data collection, validation and cleaning.
Medical
monitors assess clinical consistency and medically important findings.
Pharmacovigilance
teams manage and assess relevant safety information.
Biostatisticians evaluate
data for planned analyses and identify issues affecting interpretability.
Quality
teams provide independent assurance and support appropriate quality
processes.
Sponsors retain
responsibility for appropriate governance and oversight of trial data and
delegated activities.
Data
integrity cannot be assigned exclusively to one function.
11.
Database Lock Is a Quality Milestone
Database
lock is commonly understood as the point at which the clinical database is
finalized for planned analysis under controlled procedures.
Before
lock, the study team should confirm that relevant activities have been
completed according to the data management plan and other applicable
procedures.
These may
include:
- Resolution of critical queries
- Completion of planned data cleaning
- Appropriate medical coding
- Safety reconciliation
- Review of important protocol deviations
- External data reconciliation
- Required data reviews
- Documentation of unresolved issues
- Appropriate authorization to lock
Database
lock does not mean that every conceivable discrepancy has been eliminated.
It means
that the database has reached the defined state of readiness for its intended
analysis, with important outstanding issues appropriately addressed or
documented.
12.
Inspection Readiness Depends on Reconstructability
During an
inspection, an authority may seek to understand how a particular data point was
generated, reviewed, changed or used.
For
example:
Where did
this primary endpoint value originate?
Who entered
it?
Was it
modified?
What was
the reason?
Which
system retained the original information?
Can the
relevant records be retrieved?
If these
questions cannot be answered, confidence in the data may be reduced.
Inspection
readiness therefore depends on records and controls maintained throughout the
trial.
It should
not be created retrospectively just before an inspection.
13. Data
Governance Under ICH E6(R3)
ICH E6(R3)
places substantial emphasis on modern data governance.
Important
themes include:
- Protecting participant confidentiality
- Maintaining data integrity
- Ensuring appropriate access controls
- Defining responsibilities
- Managing computerized systems
- Maintaining audit trails
- Preserving essential records
- Applying proportionate quality controls
This
reflects the reality that clinical trial data are no longer generated and
stored in a single database.
They may
move across multiple technologies, providers and locations.
The
sponsor's governance model must account for that complexity.
14. Five
Questions Sponsors Should Ask About Data Integrity
A sponsor
should be able to answer:
1. Where do
our critical data originate?
2. How do
we know those data are accurate, complete and traceable?
3. Are the
systems generating and managing them suitable for their intended use?
4. How are
important data changes, discrepancies and transfers controlled?
5. Can we
reconstruct the data lifecycle if a regulator asks?
These
questions are more useful than simply asking whether the EDC database is
functioning.
The Agile
Clinical Trendz Perspective
At Agile
Clinical Trendz, we believe data integrity begins with good study design
and continues through site execution, monitoring, data management, medical
review, pharmacovigilance, biostatistics and quality oversight.
Reliable
clinical evidence requires:
Well-defined
data requirements
Appropriate
source documentation
Fit-for-purpose
technology
Controlled
data transfers
Effective
monitoring
Cross-functional
reconciliation
Transparent
audit trails
Clear
sponsor oversight
The
objective is not simply to produce a clean dataset.
It is to
produce a dataset whose origin, accuracy, completeness and transformation can
be understood and trusted.
Because
reliable clinical evidence does not begin with statistical analysis. It begins
with reliable data.
References
- International Council for Harmonisation. ICH
E6(R3): Guideline for Good Clinical Practice. 2025.
- U.S. Food and Drug Administration. Electronic
Source Data in Clinical Investigations. Final Guidance. September
2013.
- U.S. Food and Drug Administration. Electronic
Systems, Electronic Records, and Electronic Signatures in Clinical
Investigations: Questions and Answers. Final Guidance. October 2024.
- U.S. Food and Drug Administration. Use of
Electronic Health Record Data in Clinical Investigations. Final
Guidance. July 2018.
- U.S. Food and Drug Administration. Computerized
Systems Used in Clinical Trials. Guidance for Industry. 1999.