CCDM Clinical Trial Design 4 β Questions and Answers
Question 1: When calculating sample size for a superiority trial, increasing the desired statistical power from 80% to 90% will generally:
- Decrease the required sample size
- Increase the required sample size (Correct answer)
- Have no effect on sample size if alpha is unchanged
- Require a change in the primary endpoint
Correct answer: Increase the required sample size
Higher power means a lower probability of a Type II error (missing a true effect), which requires more participants to detect the same effect size.
Question 2: In a parallel-group randomized controlled trial, the term 'double-blind' means that:
- Neither the sponsor nor the FDA knows the treatment assignments
- Neither the participants nor the outcome assessors know the treatment assignments (Correct answer)
- The statistician and the DSMB are both blinded to interim results
- Two independent IRBs review the protocol without knowing each other's decision
Correct answer: Neither the participants nor the outcome assessors know the treatment assignments
Double-blind trials conceal treatment assignment from both participants and investigators/outcome assessors to minimize performance and detection bias.
Question 3: A 'pragmatic' clinical trial differs from an 'explanatory' trial primarily in that it:
- Tests whether an intervention works under ideal, highly controlled conditions
- Tests whether an intervention works in routine clinical practice conditions (Correct answer)
- Requires no randomization or control group
- Is conducted exclusively in academic medical centers
Correct answer: Tests whether an intervention works in routine clinical practice conditions
Pragmatic trials assess effectiveness under real-world clinical conditions with broad inclusion criteria, whereas explanatory trials maximize internal validity under controlled conditions.
Question 4: Which of the following is an example of a Type I error in a clinical trial?
- Concluding a treatment is ineffective when it actually works
- Concluding a treatment is effective when it actually has no effect (Correct answer)
- Failing to enroll enough patients to detect a real difference
- Selecting the wrong primary endpoint for the indication
Correct answer: Concluding a treatment is effective when it actually has no effect
A Type I error (false positive) occurs when the null hypothesis is incorrectly rejected, leading to the conclusion that an ineffective treatment works.
Question 5: In a cluster randomized trial, the unit of randomization is:
- Each individual patient enrolled at a site
- A group of individuals, such as a clinic, school, or community (Correct answer)
- Each investigator at participating sites
- Each individual data point collected per patient
Correct answer: A group of individuals, such as a clinic, school, or community
Cluster randomized trials randomize intact groups (clusters) rather than individuals, which is useful when individual randomization would lead to contamination.
Question 6: Which type of bias is specifically controlled by blinding in a randomized trial?
- Selection bias
- Performance bias and detection bias (Correct answer)
- Attrition bias
- Reporting bias
Correct answer: Performance bias and detection bias
Blinding controls performance bias (differential treatment of participants by caregivers) and detection bias (differential outcome assessment) by keeping group assignment unknown.
Question 7: A 'seamless Phase II/III' adaptive design is specifically intended to:
- Combine Phase I dose escalation with Phase II dose finding
- Combine dose selection or population selection with confirmatory Phase III testing in one trial (Correct answer)
- Allow switching the primary endpoint after Phase II interim analysis
- Replace Phase I entirely with a Bayesian approach
Correct answer: Combine dose selection or population selection with confirmatory Phase III testing in one trial
Seamless Phase II/III designs combine the learning (dose/population selection) and confirmatory phases into a single trial, potentially saving time and resources.
CCDM Clinical Trial Design 4
When calculating sample size for a superiority trial, increasing the desired statistical power from 80% to 90% will generally: