Data Management

in Clinical Trials

Avijit Hazra
Department of Pharmacology
IPGME&R, Kolkata

1
 Agenda

 Data management overview.

 Data management components.

Data acquisition
Data entry
Data privacy
Data storage / archival

2
 Role of data management

 Primary
role.
Ensure data integrity and reliability.

 Secondary roles.
Accelerate timeline from data collection to
analysis.
Use standard reporting and communication.
Educate users.

3
 Role of data management

Can
Can be
be achieved
achieved only
only through
through close
close cooperation
cooperation
4
 Data acquisition 1
 CRF is arguably the most important CT
document after the protocol & ICF.
 Design, development and QA of the CRF is
of utmost importance.
 Should be designed concurrently with the
protocol and may be given field trials.
 Should capture raw data rather than
calculated data.
 Should not collect redundant data and
need not capture extra data.

5
 Data acquisition 2
 Manner of data collection:
Open: Free text, numeric, alphanumeric.
Closed: Check-boxes, multiple choices.
Analog / rating scales.
 Variables must be adequately defined so that
data collected is precise and unambiguous.
 Coded formats are preferred and the coding
sequence should be as uniform as possible.
 Avoid blank items.
 Follow writing and correction protocols.

6
 Data acquisition 3
 Arrangement of data elements in the CRF should
follow the study flow but there are sections spanning
multiple visits (e.g. concomitant medication) to fill
up which reminder check boxes should be in place.
 Consider education level, intelligence, etc. if patient
is completing CRF sections (e.g. Subject Diary).
 Unique identifier, dates and signatures.
 Avoid RETROSPECTIVE filling.

7
 Data acquisition 4

Avoid RETROSPECTIVE filling.

8
 Data acquisition 5
CRF Completion Guidelines
– General instructions
– Use of permanent ink on CRFs
– List of abbreviations
– Contact information
– All items to be complete
– Procedure for making corrections on CRF
– Procedure for visits of CRFs that are not
done

9
 Data acquisition 6
Paper vs. Electronic
Electronic CRFs can/do facilitate clean data due
to functionality of application.
 Force format for data as it is entered.
 Limit responses to list of values, also force a
response so blanks are not allowed.
 Coding of AEs / Medications.
 Use of conditional logic for navigation.
 Discrepancy management features.
 BUT - Always have backup (paper) if system
fails and visit is time critical.
10
 Data entry 1
 Vital but often relatively neglected aspect
of CT data management.
 Correcting data entry errors can be costly.
 Anomalous values discovered during
analysis will bring analysis to a standstill.
 The only thing worse is failure to detect
errors, resulting in distorted results &
publication.

11
 Data entry 2
 Units & decimal points are to be treated
with utmost respect.
 Large trials allow a period of data
cleaning, verification and clarification
before analysis is commissioned.
 Data entry methods:
Single entry with no verification.
Single entry with manual verification.
Double entry with blind verification.
Double entry with interactive verification.
Double entry with independent verification.

12
Data entry 3

13
 Data privacy 1

Refers to standards surrounding

protection of personal data.
 Design CRFs with minimum number of
personal identifiers necessary.
 If source documents are to be collected
they must be stripped of personal
identifiers and appropriate coding
assigned.
 Consider a CONFIDENTIALITY
AGREEMENT for all persons with access
to data.

14
 Data privacy 2
 Enforce a NO UNAUTHORIZED
ACCESS policy and evaluate all
requests for information for a
any purposed beyond that
specified by the protocol and ICF.
 Implement procedures to ensure
privacy of data during transfer
between departments, centers,
agencies and countries.
 Appropriate security measures
for archiving of data whether
paper or electronic.

15
 Data storage 1

Data storage /
archival needs
to be
SECURE
and
EFFICIENT.

16
 Data storage 2

 Risk of data corruption and version

conflict errors increase with data
transfer operations.
 CRFs, Source documents, Raw data,
Final data sets, Database design
documentation all need to be preserved.
 Physical security must be guaranteed.

17
 Data storage 3
 Both paper and electronic data are likely
to be stored locally on-site as well as
centrally. Access control protocols are
essential.
 Electronic data can be readily backed-up
to appropriate media that can be stored
off location.
 Electronic data can be encrypted but
open format storage has its advantages.
18
 Don’t let poor data
management steal your trial !

19