Key features of clinical trial design - randomisation

Definition

Randomisation allocates trial participants to treatment groups by chance, minimising both known and unknown confounding factors by distributing them roughly equally between groups - the defining feature that distinguishes randomised controlled trials from observational studies and provides the strongest available evidence for causal inference
The methodological cornerstone of randomised controlled trials, generally positioned at or near the top of evidence hierarchies given randomisation's unique ability to balance unmeasured as well as measured confounders

Randomisation's key advantage over other allocation methods is balancing unknown/unmeasured confounders as well as known ones - observational study designs and non-random allocation methods can only adjust for confounders the researchers thought to measure, while true randomisation, given a sufficiently large sample, tends to distribute even unrecognised confounding factors roughly evenly between groups purely by chance

Allocation concealment (ensuring the person enrolling a participant cannot predict or influence which group they will be assigned to) is distinct from randomisation itself but essential to preserving its benefit - randomisation without adequate allocation concealment can still be undermined by conscious or unconscious selective enrollment patterns
Stratified randomisation (randomising within predefined subgroups, e.g. by disease severity) can be used when balancing a specific known important confounder is particularly critical, beyond what simple randomisation alone would reliably achieve in a smaller trial

A trial's randomisation and allocation concealment methodology should be specifically checked during critical appraisal, since inadequate concealment (even with technically random allocation) undermines the trial's core validity advantage
Baseline characteristic tables comparing randomised groups provide an empirical check on whether randomisation succeeded in balancing important known confounders in that specific trial

Allocation concealment as distinct from and essential to randomisation's benefit, stratified randomisation for important known confounders, baseline characteristic table as an empirical randomisation-success check, unknown confounder balancing (the unique advantage over observational designs)

Randomisation's most important advantage over careful observational study design or statistical adjustment is its ability to balance confounding factors that the researchers never even thought to measure - statistical adjustment in an observational study can only account for confounders that were actually recorded, while true randomisation tends to balance even unrecognised confounders across groups purely through the play of chance, given adequate sample size

5 more sections, plus exam facts

Premium unlocks every note across every specialty, and the full exam fact library behind it.