Cog, no trial is randomized that way.
The simplest type design would be to have cells of 3 (as this trial is 2-1) Each cell is handled by a dice roll, but will always be 2-1. With that the trial would never be out of balance by more than 2.
It is never this simple in the real world. Asnd the cell size would likely be larger, but same concept applies.
As you get more stratifcation (insuring certain groups are randomized), it can get harder to keep it in balance w/o proper design. There is plenty of math on this, and off the shelf designs are used that would have avoided this.
IMO, either:
A) Less is BS'ing
B) They used a bad design because they carried it over from a far different trial when they repurposed the P2 as a P3.
EDITL In case he is lurking, I would love to hear AVII77 weigh in.
News 
Market Data 
Discover 
