From: Choosing and evaluating randomisation methods in clinical trials: a qualitative study
Sub-theme a | Description | FG 1b (n = 6) | FG 2b (n = 5) | FG 3b (n = 9) | FG 4b (n = 5) | Totalb (n = 25) | Illustrative quote |
---|---|---|---|---|---|---|---|
Unit standard | The unit might have a specific method that would be the default – deviation from this is either rare or needs to be clearly justified | 4 | 2 | 1 | 3 | 10 | “…there’s like a default institutional sort of preference. The last place I worked that was also a default institutional preference as well for minimisation.” (Programmer 3) |
Expertise | Statisticians or programmers may be more familiar with a specific method making it easier to implement | 1 | 3 | 3 | 3 | 10 | “We have like an in-house system that we use so it doesn't really make any difference. Obviously at the start like when it was first created there was, you know, how are we going to design minimisation.” (Statistician 8) |
Cost and time | Trial set-up often has limited money and time, so resource constraints may limit the available methodology | 2 | 1 | 2 | 3 | 8 | “I guess there might be an element of cost that we need to consider because obviously minimisation would cost more” (Statistician 13) |
Trial design | The design of a trial affects how well each of the methods perform | 3 | 2 | 5 | 3 | 13 | “You look at the trial design and talk to the investigators” (Statistician 7) |
Sample size | With smaller sample sizes the affect of imbalance may be greater so more restrictive methods may be preferred | 3 | 3 | 1 | 2 | 9 | “You’re going to base it on your sample size. And also like, the number of factors that you wanted to cater for” (Statistician 15) |
Number of variables | The number of important prognostic variables you wish to include may influence this, e.g. stratification may not be able to handle a large number of variables with many strata | 3 | 2 | 3 | 1 | 9 | “I tend to question protocols when I see more than two variables and the word stratification, I think why are you doing this basically” (Programmer 2) |
External influence | The choice of randomisation method may be influenced by individuals not a member of the trial team | 2 | 3 | 2 | 0 | 7 | “But what matters is getting the publication out. And being able to show that you did something that was unpredictable and ended up with the balanced allocation in relation to key important variables. It’s the perception to the outside world, as much as anything.” (Statistician 9) |