IPPCR 2019 Issues in Randomization Part 4 of 4

>> Paul Wakim: Hello, I'm Paul Wakim. I am chief of The Biostatistics and Clinical Epidemiology Service at the NIH Clinical Center. And this is part four on randomization is clinical trials. And in this part, we're going to discuss miscellaneous points and recommendations. So, threats to the integrity of randomization. We talked about randomization methods.

But what are threats to the integrity of randomization? What can weaken the benefits of randomization? Well, exclusion of participants from final statistical analysis. Everybody got randomized, but then when it comes to the analysis, some are excluded from the analysis. Well, you don't want to do that. You really want to absolutely minimize that.

Because the more you do that, the more it basically ruins the benefit of the randomization. And same idea, if large number of participants have missing primary outcome, it's the same. It's a selection. And this is particularly concerning because you don't know whether those who are missing data have anything to do with the treatment.

So, this whole balance around randomization and objectivity and all that is basically compromised. Is this a good randomization? These are questions for you to think about. Is this a good randomization method? So, you create two groups of equal size. You, you know, 16 and 16 and then you toss a coin to determine which group is assigned to which treatment.

Is this a good randomization method? No, it is not. Why not? Because you created those two groups. And as soon as human beings are involved in the randomization process, there is potential for bias and -- well, basically for bias. Is this a good randomization method? You toss a coin to determine the first treatment assignment,