In a randomized trial, 30% of patients assigned to the active drug discontinue treatment due to side effects. The per-protocol analysis (excluding non-compliant patients) shows the drug is effective, but the intention-to-treat analysis (including all randomized patients) does not. Which analysis should be prioritized and why?
APer-protocol — it reflects the true drug effect among those who actually took it
BIntention-to-treat — it preserves the randomization that controls for confounding, even though it dilutes the treatment effect
CBoth are equally valid and should be weighted equally
DNeither — the high discontinuation rate invalidates the entire trial
Intention-to-treat (ITT) is the primary analysis in randomized trials because it preserves the validity of randomization. Excluding non-compliant patients breaks randomization — patients who tolerate the drug may differ systematically from those who do not, reintroducing confounding. The ITT analysis estimates the effect of being assigned to treatment (the pragmatic effect), which is the clinically relevant quantity: in practice, some patients will not comply. Per-protocol analysis is reported as a secondary sensitivity analysis but is not the basis for the primary conclusion.
Question 2 Multiple Choice
A double-blind trial means neither the patient nor the treating physician knows the treatment assignment. Why is blinding the physician important in addition to blinding the patient?
APhysicians who know the assignment may provide differential co-interventions, attention, or outcome assessment
BIt is only important for surgical trials, not drug trials
CIt prevents the physician from giving the active drug to sicker patients
DIt reduces the sample size needed for adequate power
An unblinded physician may unconsciously provide more careful monitoring, additional treatments, or more optimistic outcome assessments to patients in the active group — or may set a lower threshold for detecting side effects. These differential behaviors bias the comparison. Blinding the outcome assessor (triple-blinding) adds another layer of protection when outcomes involve subjective judgment (e.g., a radiologist reading scans should not know which group the patient is in).
Question 3 True / False
Randomization in a clinical trial ensures that treatment groups will be exactly identical in all baseline characteristics.
TTrue
FFalse
Answer: False
Randomization ensures that treatment groups are comparable on average across many repetitions of the randomization — it does not guarantee exact balance in any single trial. By chance, one group might have slightly older patients, more smokers, or more severe disease. This is why baseline characteristics are reported in Table 1 of every trial — to assess the balance achieved. With large samples, imbalances are typically small and do not affect conclusions. With small samples, stratified or blocked randomization can improve balance on known prognostic factors.
Question 4 Short Answer
What is clinical equipoise, and why is it an ethical requirement for conducting a randomized trial?
Think about your answer, then reveal below.
Model answer: Clinical equipoise means there is genuine uncertainty within the expert clinical community about which treatment is better. It is ethically required because randomization assigns patients to treatments they did not choose. If there were strong evidence that one treatment was superior, randomizing patients to the inferior treatment would cause avoidable harm. Equipoise justifies the experiment by establishing that the comparison is a genuine question where learning the answer benefits future patients, and neither group is being knowingly disadvantaged.
Equipoise is not about individual physician belief but collective uncertainty — even if one physician favors the new drug, the community as a whole must be uncertain. This is why interim monitoring (Data Safety Monitoring Boards) exists: if accumulating evidence during the trial shifts the balance of evidence strongly toward one arm, the trial may be stopped early to avoid continuing to randomize patients to an inferior treatment.