Ulf/Jonathan,
I would like to expand on the point I was making in my original email. There
is no such thing as a perfectly designed and executed clinical trial because
it is an attempt to carry out objective assessments of interventions in
human beings. We all know that people rarely behave the way others would
want us to.
Therefore when designing and executing a clinical trial all that can
actually be achieved is the 'best approximate fit' to the ideal scientific
design for objective measurement.
We need to bear this in mind when critically appriasing clinical trials. My
view is that there should be a two stage process that needs to be followed
before a decision is made about whether to accept or reject the results of
clinical trials.
1. Critical appraisal for possible sources of bias and other flaws.
2. Once the flaws have been identified then a question that needs to be
asked is:
'Is this trial "fit for purpose" for the question(s) it set out to ask in
spite of the limitations identified?'
If results of clinical trials are rejected without the 'fit for purpose'
assessment then there is a danger that very useful information is lost.
There is an excellent series of articles discussing this type of issues
published in the journal Circulation in 2002
(http://circ.ahajournals.org/cgi/content/full/).
David L. DeMets and Robert M Califf. Lessons Learned From Recent
Cardiovascular Clinial Trials: Part I
Circulation 2002;106;746-751
David L. DeMets and Robert M Califf. Lessons Learned From Recent
Cardiovascular Clinial Trials: Part II
Circulation 2002;106;880-886
Robert M Califf and David L DeMets. Principles From Clinical Trials Relevant
to Clinical Practice. Part I
Circulation 2002;106;1015-1021
Robert M Califf and David L DeMets. Principles From Clinical Trials Relevant
to Clinical Practice. Part II
Circulation 2002;106;1172-1175
With best wishes
Anjana
Anjana Patel BPharm, MSc, PhD, MRPharmS
Independent Pharmaceutical Scientist and Writer
UK
----- Original Message -----
From: "Underhill Jonathan" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, March 05, 2007 10:43 AM
Subject: Re: Open-labelled diabetes trial
>>>>>>>>>>>>>>>>>
As far as I am aware, the usual reason for conducting an open-label
trial is when blinding is considered to be either not practical or not
ethical.
However, in such cases bias is reduced by ensuring that the raw data is
analysed by people who are not aware of the identity of the comparative
treatments.
You mentioned trials evaluating insulin treatment in diabetes care but
do not provide information on what insulin treatment is being compared
with. In people with Type 1 diabetes it would not be ethical to withhold
insulin treatment and in Type 2 diabetes, if insulin treatment is being
compared with oral antidiabetic drugs than it would not be practical to
blind. If two different types of insulin are being compared, then
participants can usually guess which insulin they are allocated because
of having to calculate the appropriate insulin dose for food intake thus
introducing the possibility of bias and so on.
Does this answer your question?
With best wishes
Anjana Patel
<<<<<<<<<<<<<<<<<<<
Anjana/Ulf
Perhaps a more important source of bias is allocation concealment. There
are some great articles on this topic:
Schulz KF, Chalmers I, Hayes RJ, et al. Subverting randomization in
controlled trials. JAMA 1995;273:408-12.
Schulz KF, Grimes DA. Allocation concealment in randomised trials:
defending against deciphering. Lancet 2002;359:614-18.
Altman DG, Schulz KF. Concealing treatment allocation in randomised
trials. BMJ 2001;323:446-7.
Schulz KF, Grimes DA. Generation of allocation sequences in randomised
trials: change, not choice. Lancet 2002;359:515-9.
Torgerson DJ, Roberts C. Randomisation methods: concealment. BMJ
1999;319:375-6.
Summarised nicely here:
http://ebm.bmj.com/cgi/content/full/5/2/36
http://ebm.bmj.com/cgi/content/full/5/2/38
A good exercise (for those who learn by doing) is suggested here:
Turner T and Harris C. EBM 2006; 11: 5
http://ebm.bmj.com/cgi/content/full/11/1/5-a
It is important to realise that you can have allocation concealed
without it being a double-blind study. Allocation concealment can be a
much more important source of bias.
Also, a Anjana says, you can also have a study where the endpoints are
blinded (ie. The folks measuring the outcome don't know what the
participants were taking). The PROBE design (PRospective Open-label
Blinded Endpoint) is becoming more popular these days, but is not
without problems
So you can still have a non-blinded study (for the reasons given above)
that may be relatively free from bias provided the above is taken into
account.
Cheers (8(|)
Jonathan
---------------------------------------------------------------------
Jonathan Underhill MRPharmS
Head of Education and Content Development
National Prescribing Centre
The Infirmary
70 Pembroke Place
Liverpool
UK
L69 3GF
Tel: 0151-794-8143
Fax: 0151-794-8067
Mobile: 07968 851325
email: [log in to unmask]
website: www.npc.co.uk
------------------------------------------------------------------------
----
DISCLAIMER: This e-mail may contain confidential and/or proprietary
information some or all of which may be legally privileged. It is for
the intended recipient only. If any addressing or transmission error has
misdirected this e-mail, please notify the author by replying to this
e-mail and destroy any copies. If you are not the intended recipient you
must not use, disclose, distribute, copy, print, or rely on this e-mail.
|