+ Reply to Thread
Results 1 to 4 of 4

Thread: Is this really simple randomization for RCT?

  1. #1
    Points: 7, Level: 1
    Level completed: 13%, Points required for next Level: 43

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Is this really simple randomization for RCT?




    I had a debate with my friend over her study. She did a 2-arm ("A" for intervention and "B" for control) randomized controlled trial on 50 diabetic patients, recruited in sequence over 1 year's time. She did not have a treatment allocation sequence for all 50 subjects prepared in advance. Her goal was to ensure equal sample size for A & B. Here's how she allocated her subjects to groups:

    (1) Whenever a batch of 6 patients were recruited, she gave each a number, say from 1-6.
    (2) Then she used a random number generator to draw out a number. If, for instance, #5 was drawn, then this meant subject #5 was selected first.
    (3) Meanwhile, she would think of an allocation sequence for this batch, e.g. ABABAB. Accordingly, she would assign subject #5 to "A".
    (4) After #5 had been assigned, she proceeded to draw out another number and the next selected subject would be allocated to "B".
    (5) This process was repeated until all 6 subjects had been drawn out.
    (6) The sequence could be different for the next batch, such as AAABBB, since it was decided by her.

    She called her method of randomization "simple randomization" because she never knew who was the first and who was the next on the subject list.

    I thought simple randomization means EACH subject has a 50-50 probability of being allocated to A or to B, and randomization means the treatment allocation sequence, NOT the subject entry sequence, is random. Since her subjects were recruited in sequence over time, the allocation should be generated first, and individual subject would be allotted the next slot as he/she entered the study.

    Who is correct?

  2. #2
    Human
    Points: 12,676, Level: 73
    Level completed: 57%, Points required for next Level: 174
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,362
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Is this really simple randomization for RCT?

    It sounds like a randomised block to me on a first reading, where the blocks are the groups of 6 persons.

    (So it is not a completely randomized design. I would guess that the randomized block would be better in this case. Of course they must record which group/block the person belongs to.)

  3. #3
    Points: 7, Level: 1
    Level completed: 13%, Points required for next Level: 43

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Is this really simple randomization for RCT?

    Quote Originally Posted by GretaGarbo View Post
    It sounds like a randomised block to me on a first reading, where the blocks are the groups of 6 persons.

    (So it is not a completely randomized design. I would guess that the randomized block would be better in this case. Of course they must record which group/block the person belongs to.)
    tks for replying.
    It also looks like a blocked case to me. however, it is a haphazard sequence made up by her, does it still count as a 'random' sequence?

  4. #4
    Human
    Points: 12,676, Level: 73
    Level completed: 57%, Points required for next Level: 174
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,362
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Is this really simple randomization for RCT?


    Suppose that there is a linear trend in "Y" the blood level of glucose. So for every person that arrives the mean y level increases by one unit (for some reason) and that there is a difference between the persons in y. We want the randomisation procedure to protect from making erroneous conclusion about the effect of treatment A or B.

    It could be that the time trend and the patient y level could be statistically independent or being correlated.

    It seems like she both protected from the time trend by giving them numbers and protected from the individual variation by having a randomly given order order of A:s and B:s.

    I am not sure (from the reading of the text) if the blocks are of size two (so that every group of two patients they either get A or B), or if they are of size 6.


    She did a randomized controlled trial RCT, which is a wider concept than a completely randomized trial (CRT). A CRT would simply give numbers to the 50 patients and and and randomize the patient number to A or B and the time order. A CRT would not take into account of a time trend, but would protect against it by randomizing it away.

    What she did would still count as a RCD.

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats