Randomization is a draw: chance, not the surgeon, decides which treatment each patient receives. It is the only feature of a study design that balances the factors nobody thought to measure.
A single draw for every patient, like tossing a coin. Straightforward, but in small trials it can leave groups of very different sizes.
The draw runs inside blocks — of four or six, for example — so the groups stay balanced in size throughout recruitment. Block size should vary and stay unknown to the recruiting team.
Separate draws within strata that matter for the outcome (centre, stage, age band), so a prognostic factor cannot pile up in one arm.
The unit drawn is a hospital, ward or surgeon rather than a patient. It requires a larger sample and an analysis that accounts for clustering.
Yes, and ABCD publishes several. What changes is blinding: with objective outcomes and blinded assessors, a surgical trial remains valid even when patient and surgeon know the procedure.
No. If the allocation was random, any difference is due to chance by definition. Describe the groups, and adjust for prognostic variables that were defined in the protocol.
With simple randomization this is expected in small samples. Block randomization prevents it; unequal sizes alone do not invalidate the trial.
Written by Francisco Tustumi, digestive system surgeon (FMUSP), associate editor of ABCD.
Developed by Surya MKT