Randomized controlled trial

A randomized controlled trial assigns people to treatment or control by chance, so the groups differ in the treatment rather than in the kinds of people they contain. That is its entire purpose, and it is what separates it from a study that observes who already takes what.

The problem randomization solves is that people who choose a treatment differ from people who do not, in ways nobody can fully list. Those who take a supplement for longevity tend also to exercise, sleep, earn and see doctors differently. An observational study can adjust for the differences it has measured; it cannot adjust for the ones it has not. Randomization removes the need to know them, because chance spreads them evenly across both groups.

Two further safeguards usually travel with randomization, and both are worth looking for by name. Allocation concealment means whoever enrolled a participant could not know which group came next, so the sickest could not be steered away from the treatment. Blinded outcome assessment means whoever judged the result did not know which group they were judging.

They are worth looking for because they are often absent. A review of animal research reported that only about a third of studies described randomization at all, and fewer still described allocation concealment or blinded assessment. Where a safeguard goes unmentioned, a reader has no way to tell whether it was used and left out of the write-up, or never used.