A Review Found Small Variable Blocking Schemes May Not Protect Against Selection Bias in Randomized Controlled Trials

Objective: Blocking is associated with prediction of the allocation sequence and subversion. This paper explores if blocking strategies lead to an increase in baseline age heterogeneity (a marker for potential subversion) and, whether the use of blocking is changing over time.

Study Design And Settings: The British Medical Journal, Journal of the American Medical Association, The Lancet and the New England Journal of Medicine were hand searched to identify open RCTs published in January between 2001 and 2020. To explore heterogeneity of baseline age meta-analyses were performed on trials implementing blocking, minimization, and simple randomization.

Results: One hundred seventy-nine open RCTs were identified: nine (5.0%) undertook simple randomization, 104 (58.1%) blocking, 25 (13.9%) minimization, and one (0.6%) both. Baseline age heterogeneity of 24% (P= 0.02) was observed in all trials implementing blocking, 62% (P = 0.001) in trials implementing a fixed block of four, 40% (P = 0.07) implementing variable blocks including a 2 and 0% for both simple randomization and minimization. Small block sizes are implemented in modern trials.

Conclusion: Variable block sizes including two are associated with subversion and should not be implemented. If center only stratification is necessary, it should be used alongside larger blocking schemes. Authors should consider alternative methods to restrict randomization.