Hi Nina

I was recently appraising a randomised phase 2 drug trial article in a journal and i came across this 1:4 randomisation (placebo vs drug) to recruit more subjects deliberately into treatment group. The explanation of the authors was that since it is phase 2 drug trial, they wanted an 80% power to detect increased rates of adverse events. Is this randomisation valid? if NOT does this need any correction by using statistical tests?

Hello,

In response to your question, unequal randomisation is used in drug trials and can be for different reasons:

1. Sometimes you want more people in one arm: For example, if you're looking for rare events due to a drug, it may be advisable to have more people in the arm of interest, to minimise the risk of making a type 2 error.

2. Sometimes you want less people in one arm: For example, if the drug is very expensive, you may not be able to afford to have too many people recruited into the drug arm.

It is important to note that this type of randomisation method should not be used purely to recruit more people on the active ingredient and as a result produce more results for marketing purposes. A clear explanation should justify the reason for this type of randomisation method.

Also bear in mind that even under simple randomisation you could end up with a 1:4 randomisation by chance, although very unlikely. Therefore in the absence of other methodological shortcomings, it should not matter if there are more people in one arm than another. If anything, one could argue it's less natural to have equal numbers in each arm if people are being allocated by a random method. There shouldn't be a need for any statistical correction if no other factors are at play.

I hope this answers your question.

That was very helpful. Thank you very much Nina