Date:         Mon, 12 Feb 2007 10:22:34 -0500
Reply-To:     "data _null_;" <datanull@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "data _null_;" <datanull@GMAIL.COM>
Subject:      Re: Queries Regarding Randomization
Comments: To: Madan Gopal Kundu <Madan.Kundu@ranbaxy.com>
In-Reply-To:  <77943F39FD8BFB459C22A7A2E578626A059CCCA8@NCORP-MSG-CLS.Ranbaxy.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Seems to me that if the treatment ratio is 2:2:1 then block size
should be five.  In a clinical trial a BLOCK is about a far from the
classical definition of BLOCK as you can get.

A BLOCK should consist of a group of homogeneous experimental
material, litter mates, areas of a field, units chosen with similar
qualities specific to the experimental material.

This is almost never accomplished in a clinical trial nor is it even
attempted.  People are too variable for this type of blocking.  Unless
you use twins, or the subjects are blocks as in the cross over design.

Clinical trials are bocked because of sequential enrollment.  You want
to try and enroll a complete block of subjects according the the
ratio.  This way, if you have very low enrollment at a center chances
are it will have at least one complete set of treatments in the chosen
ratio.

Therefore to set a block size of 9 for a 2:2:1 treatment ratio seems
illogical.  What would be the point.

This is assuming the common parallel design which is indicated by your
PROC PLAN statements.


On 2/11/07, Madan Gopal Kundu <Madan.Kundu@ranbaxy.com> wrote:
> Dear All,
>
> I have query regarding randomization of clinical trials.They are:
>
> 1. Consider the following situation: block size is 9 and treatment
> allocation ratio of three treatments is 2:2:1. More generally, block
> size is NOT divisible by sum of the treatment allocation ratios  --- Can
> this kind of situation appear?
>
> 2. If the above kind of situations appear then how to generate
> randomization schedule? I mean when block size (say 10) is divisible by
> treatment allocation ratio (say, 2:2:1) then we can generate
> randomization schedule in the following way:
>
> proc plan seed=123456;
> factors block=15 trt=10;
> output out=randout
>       trt cvals=('A' 'A' 'A' 'A' 'B' 'B' 'B' 'B' 'C' 'C') random;
> run;
>
> But when block size is 9 and treatment allocation ratio is 2:2:1 then
> how to generate randomization schedule.
>
> I am looking forward to get your comments.
>
> Thanks and Regards
> Madan Gopal Kundu
>