Randomized Controlled Trials and Experimental Designs: Analysis, Randomization, and Reporting

This chapter introduces the methodological and statistical foundations of randomized controlled trials (RCTs) in epidemiological research. The chapter explains why RCTs are widely regarded as the gold standard for causal inference, emphasizing how random allocation to intervention and control groups helps reduce selection bias and balance measured and unmeasured confounding factors. Students learn to distinguish statistical association from causation and understand the Average Treatment Effect (ATE) as the difference between the expected outcomes under intervention and control conditions. The material also examines three randomization strategies, simple, blocked, and stratified, and explains their advantages and limitations for achieving balanced treatment groups. Through a simulated lifestyle intervention study examining changes in systolic blood pressure, students explore how statistical comparisons, confidence intervals, hypothesis testing, and graphical methods can be used to evaluate intervention effects. The chapter further emphasizes that strong internal validity does not automatically guarantee generalizability and that ethical principles, including clinical equipoise and informed consent, are essential to conducting RCTs responsibly.

The second major focus is estimands, analytical strategies, and practical challenges in randomized trials. Students examine the distinction between intention-to-treat (ITT) and per-protocol (PP) analyses. ITT retains participants in their originally assigned groups, preserving the benefits of randomization and estimating the effect of treatment assignment, whereas PP focuses on participants who comply with the study protocol and may be more vulnerable to selection bias. The chapter also addresses noncompliance, participant dropout, and missing data, explaining how these problems can affect treatment-effect estimates, statistical precision, and study validity. It also introduces methods such as multiple imputation to address missing observations, along with the importance of preventing missing data through appropriate study planning and participant-retention strategies. Students also learn the principles of sample size determination and statistical power, including the roles of significance level, type II error, expected effect size, variability, and anticipated dropout. The material shows how R can calculate sample sizes for continuous and binary outcomes and explains why sample size planning must balance statistical precision, clinical relevance, resource efficiency, and ethical responsibility.

The final component focuses on outcome-type-specific statistical analysis and transparent reporting of RCT findings. Students learn to select appropriate methods for binary, continuous, and time-to-event outcomes. For binary outcomes, the chapter covers risk difference (RD), risk ratio (RR), odds ratio (OR), chi-square and Fisher’s exact tests, and logistic regression. For continuous outcomes, it discusses independent-samples t-tests, the Mann–Whitney test, mean differences, standardized mean differences, and analysis of covariance (ANCOVA) to improve precision by accounting for baseline measurements. For time-to-event outcomes, the material introduces Kaplan–Meier estimation, the log-rank test, and Cox proportional hazards regression, including hazard ratio interpretation and the proportional hazards assumption. The chapter concludes with the CONSORT reporting guidelines, emphasizing transparent descriptions of randomization, allocation concealment, blinding, participant flow, sample size calculations, outcome definitions, effect estimates, 95% confidence intervals, and adverse events. It also highlights trial registration and consistency between the original protocol and published results to reduce selective reporting and improve reproducibility. Overall, this chapter equips Statistics students with the knowledge and analytical skills needed to design, analyze, critically evaluate, and transparently report experimental epidemiological studies that produce credible evidence for clinical and public health decision-making.