Consort statement portrays minimisation as follows:
"Minimization ensures balance between intervention groups for several
patient factors. Randomization lists are not set up in advance. Although the
first patient is truly randomly allocated; for each subsequent patient, the
treatment allocation is identified which minimizes the imbalance between
groups at that time. That allocation may then be used, or a choice may be
made at random with a heavy weighting in favor of the intervention that
would minimize imbalance (for example, with a probability of 0.8). The use
of a random component is generally preferable. Minimization has the
advantage of making small groups closely similar in terms of participant
characteristics at all stages of the trial. Minimization offers the only
acceptable alternative to randomization, and some have argued that it is
superior. Trials that use minimization are considered methodologically
equivalent to randomized trials, even when a random element is not
incorporated."
http://www.consort-statement.org/examples8b.htm
Mustafa Soomro
Hon. Research Fellow
St. Goerges Hospital Medical School, London
-----Original Message-----
From: Evidence based health (EBH)
[mailto:[log in to unmask]]On Behalf Of Andrew Jull
Sent: 06 November 2005 21:03
To: [log in to unmask]
Subject: Re: [EVIDENCE-BASED-HEALTH] Is this study randomized?
Interesting question
There was an editorial regarding minimisation in BMJ that argued
minimisation is a better method of ensuring group equivalence across a range
of a priori determined prognostic factors than stratified randomisation
(Treasure T, MacRae KD. BMJ 1998;317:362-363). In discussions I have had
with our senior statistician, I know that there are people who believe that
minimisation is not randomisation. However, I have always taken it to be
something of an advanced form of randomisation as a minimisation algorithm
involves both the play of chance and a check that the allocation is not
creating an imbalance between groups - or that is my rudimentary
understanding of the method - I would welcome further clarification.
Regards
Andrew Jull
Research Fellow
Clinical Trials Research Unit
University of Auckland
-----Original Message-----
From: Brian Alper MD [mailto:[log in to unmask]]
Sent: Monday, 7 November 2005 12:51 a.m.
To: [log in to unmask]
Subject: Is this study randomized?
In applying the SORT criteria to a placebo-controlled trial of infliximab
for psoriasis (Lancet 2005 Oct 15;366(9494):1367), all quality criteria for
level 1 evidence appear to be met except randomization. The study does not
state the word random or randomization anywhere.
The study states:
An adaptive treatment allocation (stratified by investigator site) was used
to assign patients. Eligible patients were allocated in a 4:1 ratio to
receive infliximab 5 mg/kg (Remicade, Centocor, Malvern, PA, USA) or
placebo. ClinPhone (Lawrenceville, NJ, USA) allocated patients using a
minimisation algorithm with biased coin assignment by means of an
interactive voice response system. The treatment assignment was stored
electronically, and the stored data were used to allocate future patients in
such a way that the imbalance between treatment groups was kept to a
minimum. This information was distributed to the unblinded site pharmacist,
who then prepared the study drug. The investigators, study site personnel,
and patients remained blinded until the database lock at week 50.
My question is whether this is a sophisticated form of stratified
randomization and should be considered level 1 evidence or whether this
should be considered a non-random design.
|