Topic 10: Practical Considerations in Clinical Trials

Question 1

Who decides whether a trial should stop or continue when doing interim analysis

Answer 1

Data monitoring and ethics committee (DMEC)

Question 2

What do the DMEC use to guide the stop/go decision

Answer 2

Efficacy, futility, safety, quality.

Question 3

When should the number and nature of any interim analyses be decided

Answer 3

The design stage

Question 4

Give 2 examples of assumptions that may be checked at the interim analyses

Answer 4

Control arm event rates, variance of primary outcome, drop out

Question 5

True/False: Sample size can be reduced following an interim analysis

Answer 5

False. Sample size can be increased if necessary but must not be reduced.

Question 6

What is a sequential trial

Answer 6

A type of adaptive trial design using interim analyses. Looks at data after each pair of partcipants

Question 7

What is used instead of a sequential trial and why

Answer 7

Group sequential trial, because participants seldom arise in pairs, follow up takes time, very inefficient to download and analyse data after every pair of participants.

Question 8

What is a group sequential trial

Answer 8

A trial that analyses results at various intervals, after a pre-specified number of individuals (2n) or by date, with the intention of stopping the trial early according to pre-specified rules, when a conclusive result has been reached.

Question 9

What is nominal type 1 error

Answer 9

The value of the split up alpha that is used to control for type 1 error. So if using Bonferroni correction, nominal error would be alpha/n.

Question 10

Why is it hard to choose nominal type one error when using repeated tests

Answer 10

The tests are highly correlated

Question 11

Which correlation between consecutive tests is greater: sequential trials or group sequential trials

Answer 11

Sequential

Question 12

What is a type of group sequential trial

Answer 12

Double triangular trial.

Question 13

Give 3 kinds of stopping rule for interim analysis

Answer 13

Pocock, O’Brien and Fleming, Peto

Question 14

How does the Pocock stopping rule work

Answer 14

Use the same nominal type 1 error for all tests

Question 15

Why is Pocock rule unpopular

Answer 15

Likely to stop the trial early

Question 16

What does O’Brien and Fleming stopping rule say

Answer 16

Increase the nominal type 1 error rate as stage increases and this allows the final significance level to be closer to the overall type 1 error

Question 17

What does the Peto Stopping Rule Say

Answer 17

Have very low nominal type 1 error for all tests apart from the final tests, then desired type 1 error for final test - this assumes the alpha used at interim analyses is negligible.