Section III: Experimental Design & Randomized Controlled Trials

Randomized Controlled Trial (RCT)

An experimental study where participants are randomly assigned to an intervention or control group to test a causal hypothesis.

Randomization

The process of assigning participants to groups by chance to eliminate selection bias and balance confounders.

Purpose of Randomization

Ensures comparability between groups on known and unknown confounders; supports causal inference.

Allocation Concealment

Hiding upcoming group assignments from recruiters to prevent selection bias before randomization.

Blinding

Keeping participants, clinicians, or analysts unaware of assignment to prevent measurement bias.

Single vs Double Blinding

Single = participant unaware; Double = participant and investigator unaware.

Control Group

The comparison group not receiving the experimental intervention, used to estimate counterfactual risk.

Internal Validity

Accuracy of results within the study population; threatened by loss to follow-up or bias.

External Validity

Extent to which study findings generalize to other populations or settings.

Confounding

A third variable related to both exposure and outcome that distorts the true association.

Adherence

Degree to which participants follow assigned protocol; non-adherence can bias toward the null.

Intention-to-Treat (ITT) Analysis

Analyzes participants according to their original randomization, preserving benefits of randomization.

Per-Protocol Analysis

Analyzes only adherent participants, showing biological efficacy but risking bias.

Cluster RCT

Randomization by group or site (e.g., clinic, village) instead of individuals.

Crossover Trial

Each participant receives both treatments sequentially, separated by a washout period.

Factorial Trial

Tests two or more interventions simultaneously using multiple randomization factors.

Phase I Trial

Tests safety, dosage, and side effects (20–80 participants).

Phase II Trial

Tests preliminary efficacy and safety (100–300 participants).

Phase III Trial

Confirms efficacy vs standard, monitors adverse events (1 000–3 000 participants).

Phase IV Trial

Post-marketing surveillance for long-term or rare effects.

Placebo Effect

Improvement due to expectation rather than active treatment.

Placebo Controlled means?

one group receives a new, active treatment and a control group receives a placebo, an inactive substance designed to look, taste, and smell like the real treatment

Hawthorne Effect

Behavior change because participants know they’re being observed.

Salk Polio Vaccine Trial

Large double-blind placebo-controlled RCT demonstrating vaccine efficacy; landmark in public health.

Internal vs External Validity

Internal = trust your study’s findings; External = apply them elsewhere.

Selection Bias in RCT

Occurs if randomization or allocation concealment fails and group entry differs systematically.

Loss to Follow-Up

Reduces internal validity if drop-outs differ by exposure or outcome

Survival Analysis

Statistical methods for analyzing time until an event occurs (e.g., death, relapse).

Event '

The outcome of interest (e.g., death, relapse, infection).

Censoring

Participant’s follow-up ends before event occurs (lost to follow-up, study end, withdrawal).

Survival Function S(t)

Probability of surviving beyond time t; starts at 1 and decreases over time.

Kaplan–Meier (KM) Estimator

Non-parametric method estimating S(t) with step decreases at each event time.

Censored Observation on KM Curve

Shown as a tick mark; does not cause a drop but reduces future sample size

Life Table

Interval-based survival method using conditional probabilities for each time interval.

Median Survival Time

Time when S(t)=0.5; half the cohort has experienced the event.

Hazard Ratio (HR)

Ratio of hazards between two groups (h1/h0). HR < 1 = protective; HR > 1 = increased risk.

Log-Rank Test

Statistical test comparing survival curves between groups