who tends to make most of the mistakes. Is CDC
review adequate? Are the programs written for
the computer edit checks sufficient? Did the
data entry verifier find the problems of the
keyer and fixed them? Once the CDC review
manual and computer edit checks have been
improved, a second interim audit should be
repeated when 50–60% of the CRFs have been
scanned and entered. Final audit should be per-
formed when all the CRFs are scanned and
entered. In addition to the audit memo issued
during the interim audits, an audit summary
report (Table 26.10) should be issued to sum-
marize the quality of the final database. This
will also give management an index of the
error rate, which measuresthe confidence level
of releasing the database.
Database release memo
Once all the queries have been resolved and
updated to the image files and database, the data-
base is officially locked. A database release memo
(Table 26.11) should be issued to the project sta-
tistician, all other team members and management.
The project statistician will then merge the file of
therandomizationcodes tothe database to generate
the analysis datasets.26.5 Disaster recovery plan
The data files and the completed CRFs generated
from the clinical trials are more precious than
the hardware. In addition to daily and monthly
back-up, pharmaceutical companies should have
a detailed recovery plan in case of unexpected
disaster. The disaster recovery plan should include
the following:1.Key personnel contact list, with home telephone
and pager numbers listed.2.List of critical applications and operations.
It should be a company’s policy to set up an off-
site processing center with the same hardware–
software setup. This should be able to be made
operational within 2 h should it become necessary
if a disaster strikes the data center. Critical appli-
cations and operations include upcoming NDA
studies, safety database, NDA summaries.3.Off-site storage. In order to be operational at the
off-site process center for critical applications,
the files that need to be updated to the off-site
center are master files for all completed projects,
daily back-ups for ongoing studies, monthly
system files, monthly glossary updates, monthly
safety monitoring, NDA files and production job
streams. The protocols, CRFs, regulatory docu-
ments, rules and manuals should also be stored
off-site. A drill of the disaster recovery plan
should be put to test at least every 6 months to
reveal any unanticipated problem.References
ICH. 1996. ‘International Conference on Harmoniza-
tion: Guideline on Structure, Content and Clinical
Study Reports; availability; notice’.Federal Regis-
ter 61 (138): 37326.
PhRMA. 1997.Facts and Figures.Table 26.11 Database release memorandum
Date:
To: Project statistician
From: CDC
Subject: Database Release
The master files of the following study have been
audited by the QA department and passed. The
database is therefore signed off and released by
Data Management.
Drug:
Protocol number:
Investigator number and name:
Date master files signed off:
Time master files signed off:
Please verify that you have received the master files.
Via this memo, QA department is requested to
release the randomization codes to the project
statistician. Thanks.
Clinical Data coordinator
cc: Project team members
354 CH26 DATA MANAGEMENT