It may go without saying that we design our clinical trials in a manner that will achieve our objectives while requiring the least amount of resources. It is less obvious just how well statistics can help us do it.
There are numerous bio statistical methods that can make your trial more efficient, offering considerable savings of both time and money. Following are a few examples of how to assess the most fitting primary efficacy endpoint using bio statistical methods:
The ICH E9 guideline states:
“The primary variable (‘target’ variable, primary endpoint) should be the variable capable of providing the most clinically relevant and convincing evidence directly related to the primary objective of the trial.”
In many indications the primary efficacy endpoint is well established and you will be constrained to select the standard endpoint. Yet even in these cases you might have some control over your design to make it most efficient. Specifically, it would behoove you to ask and answer the following questions:
- How do I best measure the primary endpoint? (Use a single central lab, reviewer or committee where appropriate and feasible)
- When should I measure my endpoint? (Analyze historical data—your own and others’—to find the time point that yields the largest difference between test and control.)
- How many times should I measure the endpoint within a time point? Across time points? (To the degree possible, use multiple measurements and analyze accordingly. The number of repeated measurements taken will depend on both statistical and clinical/logistic considerations.)
In cases where you have more control over endpoint selection the following considerations will also be important and have been demonstrated to reduce sample size by 10% to 50% and greatly increase power for statistical testing:
- Avoid responder (response/non-response) endpoints in favor of continuous endpoints such as size, mmHg, weight, etc.
- Prefer continuous endpoints to rank endpoints such as NYHA functional classification.
- When obtaining patient reported outcome (PRO), select the validated scale with the greatest reliability (repeatability).
- To the degree possible, analyze your efficacy endpoint using procedures that are robust to missing data, such as mixed model repeated measures (MMRM).
Additional methods for increasing your chance for success in a trial with fewer resources include:
- Adaptive designs
- Minimizing variance
- Specifying matched subject designs
- Designing single arm trials for regulatory approval
- Making efficient use of non-inferiority hypotheses
- Reducing noise by both appropriate measurement and statistical analysis
All of the methods described above, used in combination or alone, will increase your trial’s efficiency by reducing costs and increasing probability of success.