Basic concepts and statistical aspects

Summary

Clinical study designs: key concepts

Clinical studies are generally divided into observational and experimental designs.

• Observational studies (e.g., cohort, case-control, cross-sectional) assess associations without active intervention. While essential for hypothesis generation and studying rare exposures, they are considered weaker in terms of causal inference compared with experimental designs due to susceptibility to confounding and bias.
• Interventional studies, particularly the randomized controlled trial (RCT), are considered the gold standard for evaluating the efficacy and safety of interventions.

Types of RCT designs

• Inter-patient (parallel group) designs: Participants are randomized to different intervention arms, studied concurrently. This design minimizes carry-over effects but generally requires a larger sample size.
• Intra-patient (crossover) designs: Participants receive treatments sequentially, serving as their own controls. This design reduces inter-individual variability but risks carry-over effects, where the first treatment influences outcomes of the second.

Sources of error in clinical trials

The observed treatment effect may be influenced by:

1. Systematic errors (bias):
   RCTs aim to minimize bias through randomization, double blinding, control groups, and intention-to-treat (ITT) analysis.
2. Random (chance) errors:
   Random variability can obscure true effects. This is mitigated by increasing sample size, which improves statistical power.

Control Groups

Control arms may involve:

• Placebo (negative control): No active intervention.
• Best available treatment (positive control): Active comparator to assess relative efficacy.

Endpoints and Blinding

• Endpoints: Primary and secondary endpoints must be clearly defined in the trial protocol to ensure reliable interpretation and regulatory compliance.
• Double blinding: Blinding of both investigators and participants reduces observer and patient expectation bias, improving internal validity.

Analysis Populations

Trial analysis may follow two main approaches:

• Per protocol (PP): Includes only participants who completed the study according to protocol. This may overestimate efficacy.
• Intention to treat (ITT): Includes all randomized participants, regardless of protocol adherence. ITT reflects real-world clinical effectiveness and is generally preferred.