Introduction
Biostatistics principles for clinical trials MCQs With Answer is a concise question bank designed for M.Pharm students preparing for exams in Clinical Research & Regulatory Requirements. This collection focuses on essential statistical concepts applied to clinical trials — hypothesis testing, errors and power, randomization strategies, survival analysis, sample size determination, handling multiplicity and missing data, interim analyses, and trial design types (non-inferiority, equivalence, crossover, cluster). Each multiple-choice question is crafted to reinforce both conceptual understanding and practical application in study design, analysis, and regulatory interpretation. Answers are provided for quick self-assessment and targeted revision to build competence for academic and professional roles in clinical research.
Q1. What does a p-value represent in the context of hypothesis testing in clinical trials?
- The probability that the null hypothesis is true given the observed data
- The probability of observing the data, or something more extreme, if the null hypothesis is true
- The probability that the alternative hypothesis is true given the observed data
- The probability of making a Type II error
Correct Answer: The probability of observing the data, or something more extreme, if the null hypothesis is true
Q2. In a clinical trial, a Type I error (alpha) is best described as:
- Failing to detect a true treatment effect
- Rejecting the null hypothesis when it is actually true
- Accepting the null hypothesis when the alternative is true
- Overestimating the treatment effect size
Correct Answer: Rejecting the null hypothesis when it is actually true
Q3. Statistical power of a trial is defined as:
- The probability of making a Type I error
- The probability of correctly rejecting the null hypothesis when the alternative is true
- The proportion of missing data that can be tolerated
- The expected p-value under the alternative hypothesis
Correct Answer: The probability of correctly rejecting the null hypothesis when the alternative is true
Q4. Which statement correctly distinguishes a two-sided test from a one-sided test?
- A two-sided test only detects increases in the primary endpoint
- A two-sided test assesses for a difference in either direction from the null value
- A one-sided test requires a larger sample size than a two-sided test for the same power
- A one-sided test evaluates equivalence while a two-sided test evaluates non-inferiority
Correct Answer: A two-sided test assesses for a difference in either direction from the null value
Q5. The intention-to-treat (ITT) principle in randomized clinical trials mandates:
- Excluding any subject who deviates from study protocol from the final analysis
- Analyzing participants according to the treatment actually received
- Including all randomized participants in the analysis according to their assigned groups
- Pooling results from different trials without adjustment
Correct Answer: Including all randomized participants in the analysis according to their assigned groups
Q6. Allocation concealment in randomization primarily aims to prevent:
- Performance bias after treatment allocation is revealed
- Bias in outcome assessment by blinded investigators
- Selection bias arising from foreknowledge of upcoming assignments
- Measurement error in laboratory assays
Correct Answer: Selection bias arising from foreknowledge of upcoming assignments
Q7. Which randomization method is optimal to ensure balance across important baseline prognostic factors?
- Simple randomization without restrictions
- Block randomization with a fixed block size only
- Stratified randomization with separate randomization lists within strata
- Cluster randomization by treatment center
Correct Answer: Stratified randomization with separate randomization lists within strata
Q8. The Kaplan-Meier estimator is used to:
- Estimate the mean of a normally distributed continuous endpoint
- Estimate the survival function or time-to-event probability while accounting for censoring
- Compare variances between two independent groups
- Adjust for multiple comparisons in interim analyses
Correct Answer: Estimate the survival function or time-to-event probability while accounting for censoring
Q9. The log-rank test in survival analysis is primarily used to:
- Adjust hazard ratios for covariates in a multivariable model
- Compare survival distributions between two or more groups
- Estimate median survival time for a single group
- Test proportional hazards assumption
Correct Answer: Compare survival distributions between two or more groups
Q10. A key assumption of the Cox proportional hazards model is that:
- Baseline hazard functions are identical across groups
- Hazard ratios between groups remain constant over time
- Event times follow a normal distribution
- There is no censoring of outcome data
Correct Answer: Hazard ratios between groups remain constant over time
Q11. In a non-inferiority clinical trial, the non-inferiority margin represents:
- The expected benefit of the new treatment over placebo
- The maximum clinically acceptable difference by which the new treatment may be worse than active control
- The minimum sample size required to demonstrate superiority
- The alpha spending function for interim looks
Correct Answer: The maximum clinically acceptable difference by which the new treatment may be worse than active control
Q12. An equivalence trial differs from a non-inferiority trial in that an equivalence trial:
- Aims to show the new treatment is superior to standard of care
- Uses a one-sided confidence interval to declare success
- Aims to show the treatment effect lies within pre-specified upper and lower bounds
- Does not require pre-specification of margins
Correct Answer: Aims to show the treatment effect lies within pre-specified upper and lower bounds
Q13. Interim analyses with potential early stopping for efficacy should use pre-specified boundaries because:
- They increase Type II error without adjustment
- Unplanned looks cannot affect Type I error
- They control overall Type I error inflation due to multiple looks at data
- They remove the need for final analysis
Correct Answer: They control overall Type I error inflation due to multiple looks at data
Q14. Multiplicity in clinical trials (multiple endpoints or comparisons) primarily increases the risk of:
- Type II error (false negatives)
- Type I error (false positives)
- Loss to follow-up
- Violation of randomization
Correct Answer: Type I error (false positives)
Q15. A 95% confidence interval for a treatment effect should be interpreted as:
- The probability that the true effect lies within this interval is 95%
- The treatment effect will be clinically significant if the interval excludes zero regardless of context
- The interval contains values that are compatible with the observed data under repeated sampling
- The p-value is guaranteed to be less than 0.05 if zero is outside the interval
Correct Answer: The interval contains values that are compatible with the observed data under repeated sampling
Q16. Which change will generally increase the required sample size for a superiority trial, holding other factors constant?
- Expecting a larger effect size between treatments
- Accepting a higher Type I error (larger alpha)
- Requiring higher statistical power (e.g., from 80% to 90%)
- Using a more liberal endpoint definition that decreases variability
Correct Answer: Requiring higher statistical power (e.g., from 80% to 90%)
Q17. When data are missing at random (MAR), which approach is generally preferred for handling missing outcome data?
- Complete-case analysis only using subjects with full data
- Last observation carried forward without modeling
- Multiple imputation that accounts for uncertainty in missing values
- Dropping the variable with missing data from analysis
Correct Answer: Multiple imputation that accounts for uncertainty in missing values
Q18. A per-protocol analysis in a randomized trial refers to:
- Analyzing all randomized participants regardless of adherence
- Analyzing only participants who completed the study according to the protocol
- Pooling data across trials in a meta-analysis
- Using intention-to-treat principles with additional imputation
Correct Answer: Analyzing only participants who completed the study according to the protocol
Q19. For cluster-randomized trials, the intra-cluster correlation coefficient (ICC) is important because it:
- Indicates the individual subject-level variance only
- Determines the degree of similarity of outcomes within clusters and affects effective sample size
- Can be ignored if cluster sizes are equal
- Directly replaces the need to randomize clusters
Correct Answer: Determines the degree of similarity of outcomes within clusters and affects effective sample size
Q20. The area under the ROC curve (AUC) for a diagnostic test represents:
- The proportion of true positives among all positive test results
- The probability that a randomly chosen diseased subject has a higher test value than a randomly chosen non-diseased subject
- The optimal cut-off value for the test
- The sensitivity multiplied by specificity
Correct Answer: The probability that a randomly chosen diseased subject has a higher test value than a randomly chosen non-diseased subject
