[EpiData-list] Definition of GCP guidelines and tools

epidata-list at lists.umanitoba.ca epidata-list at lists.umanitoba.ca
Sat Jul 8 18:44:31 CDT 2006


epidata-list at lists.umanitoba.ca wrote:
> Tim C. Responded: Which set(s) of GCP guidelines is Epidata using as the
> goal?
> 
> GCP is an abbreviation of "Good Clinical Practice".
> 
> The whole issue of GCP guidelines is a difficult one. The preparation in
> itself involves reading a lot of " heavy legal text". The major issue is
> that the principles are described in lengthy documents, but there is no
> strict definition as of what it really implies in terms of specific
> implementation (apart from the principles). I will add a summary of the
> main documents later.
> 
> In the most recent time I have been experimenting with various software
> tools for the task:
> - bug tracking
> - developmental documentation of tasks resolved.
> - documentation in general
> - testing and verifying software behaviour
> - how to accomplish solutions which are operating system independent.
> ........
> 
> To develop all the details of this is going to be a lot of work, which
> is why collaborative efforts are crucial. One of the issues is
> persontime and funding and another is testing. As soon as I can reveal
> more details of the work plan and steps of this process the list will be
> informed. And interested persons invited to participate.

Ah, OK, I misunderstood - you are referring to adapting GCP principles
to the software engineering of EpiData, rather than ensuring that
EpiData meets the needs of those who are trying to practice GCP in their
clinical work. Regarding the second aim, I think that a) those
requirements cannot easily be described in a single document or set of
requirements - it depends so much on the nature of the clinical work and
the type of data and study designs being used, and no one software tool
can every hope to meet all those requirements; and b) for many tasks
within the scope of data collection and management, and exploratory data
analysis, EpiData does a good job already.

Regarding the first goal, I wonder whether GCP documentation is the best
place to look? There is a large literature on software quality and
quality assurance in software engineering in the computer science
literature - it might be quicker to consult that body of knowledge
rather than trying to glean principles from GCP guidelines. I doubt that
you will find much discussion of things like formal test plans,
calculation of the coverage of unit tests, and the issues of testing
"corner cases" and how to check for "numerical bruising" and other
subtle computational problems in the GCP guidelines, but those topics
are pretty standard in most texts on quality assurance and testing in
software engineering - or so I am told. I have just been reading some
lecture notes from some undergraduate computer science courses on these
topics and found them very instructive and helpful indeed.

Tim C




More information about the EpiData-list mailing list