Simple randomization did not protect against bias in smaller trials.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Evidence-Based Medicine

Background:

• Simple randomization aims to prevent bias by ensuring equal treatment group distribution.
• Insufficient sample sizes in randomized controlled trials can lead to residual random differences, potentially biasing treatment effect estimates.

Purpose of the Study:

• To determine the minimum sample size necessary to eliminate random differences in treatment groups.
• To ensure unbiased estimation of treatment effects in clinical trials.

Main Methods:

• Reanalysis of two large, simple, multicenter trials (International Stroke Trial and CORONARY) with 1,000 iterative analyses on varying sample sizes.
• Monte Carlo simulations of randomized trials were conducted to supplement empirical findings.
• Assessment of covariate balance and estimation of treatment effects (aspirin/heparin for stroke, off-pump CABG for cardiovascular outcomes).

Main Results:

• Randomization effectively removes random differences and minimizes bias with at least 1,000 participants.
• Substantial bias emerges in analyses with fewer than 1,000 participants.
• Approximately 41.5% of phase 3 trials in high-impact journals achieve this enrollment threshold.

Conclusions:

• Conclusions from small randomized trials may lack reliability due to potential bias.
• Consideration of alternative allocation strategies, such as minimization or blocking, is recommended for trials with limited participant numbers.