I get the impression that the stratification factors can impact the final randomization outcome. The goal might be a 2 to 1 ratio per clinical site, but this would never be exactly 66.7% - 33.3% (for example if a hospital has a total of 4 or 5 patients, then it will never be a 2 to 1 ratio) ... and with many, many sites this could throw of the final cumulative percentages?