Principles and Practice of Pharmaceutical Medicine

documentation of IMP shipments, accountability,
reconciliation and destruction; and randomization
code break envelopes to determine that they are
complete and intact. Any code breaking must be
fully documented.
Verification of informed consent formsfor all
trial participants is a key task during audits. The
auditor should check if an informed consent form is
present for all trial subjects and has been signed by
the subject and the investigator prior to any trial-
related activity.
A major component of investigator site audits is
devoted to verify the validity of the clinical data
generated and recorded at the investigator site. This
step includes theaudit of a sample of CRFsagainst
source documents and original medical records.
The purpose of the review is to determine if the
trial procedures followed at the site are complying
with protocol requirements, if the data gathered is
complete and accurately transcribed onto the CRF
or electronic forms and if the clinical monitoring
andSDVprocess issatisfactory(FDAGuidancefor
Industry, 1988; DGGF, 2003). If computerized
systemsare used at the site to capture data, these
should also be reviewed to ensure security, retrie-
vability and validity (FDA 21 CFR Part 11; FDA
Guidance for Industry, 1999).
The investigator site audit concludes with a
closing meeting with the investigator and key site
personnel to review key audit findings and to sug-
gest corrective and preventive action, if required.

Database audit

Following collection of the CRFs from the inves-
tigator sites, the clinical data are transcribed or
transmitted on to an electronic database. Data
entry and verification, data cleaning and consis-
tency checks and coding of medical terminology
such as adverse events, concomitant medication
and medical history are procedures which are
prone to error. Therefore, periodic checks, in-
process quality control steps, should be implemen-
ted in the data management process. An audit of the
database by QA helps ensure that data integrity and
validity have not been impaired during data man-
agement procedures.

Clear procedures (SOPs) for conducting such

audits must be established, detailing the sampling

procedures for CRFs and acceptable error rates.

Information is available in literature on error levels

and data verification procedures (DGGF, 2003;

Zhang, 2004; Society of Clinical Data Manage-

ment, 2005).

For the database audit to be meaningful, the

database should only be audited in a ‘frozen’ or

‘defined’ state and prior to database lock so that

eventual changes are possible after the audit with-

out requiring a database ‘unfreeze’. When compar-

ing CRFs and data queries against the database

entries, data entry, data validation and coding pro-

cedures should be taken into account. It is impor-

tant to ensure that no changes were made to the

clinical data without proper justification and com-

plete documentation. Depending on the number

and volume of CRFs to be verified, database audits

can be quite time-consuming.

Report audit

The study report is the essence of the clinical trial

and summarizes trial data and their interpretation.

Since trial reports are part of the package submitted

to regulatory authorities for obtaining marketing

authorization, the contents must be valid, complete

and accurate. Trial report audits verify that all

necessary components and attachments are

included in the report. Ideally, the last draft version

is subject to audit, thus avoiding rework which may

be necessary after audits of early drafts which are

substantially changed until they are considered

final. In addition, all QC checks and activities

should have been completed prior to the audit.

Apart from compliance with SOPs for biostatis-

tics and report writing, the statistical analysis plan,

the trial protocol, regulatory requirements and

guidelines (ICH E3, 1995; ICH E9, 1998; ISO

9000:2005, 2005), QA auditors check the internal

consistency of the trial report and appendices and

between data in tables, figures and graphs and num-

bers cited in the text. All numbers and percentages

mustbesubstantiatedbyattachedtablesandlistings.

In summary, the trial report should be an accurate

representationoftheclinicaldata.Allocationoftrial

13.6 BRIEF OUTLINE OF AUDIT TYPES 171