Randomised controlled trials may have many unrecognised potential biases

Randomised controlled trials (RCTs) may be the gold standard, but all that glisters is not gold, warns a research fellow at the London School of Economics.1

Alexander Krauss has analysed the 10 most commonly cited RCTs in the medical literature—all with 6500 or more citations—and found that they have potential biases that are frequently unrecognised or unacknowledged by the trials’ originators. The claim that randomisation eliminates bias is simply not true, he says in Annals of Medicine.1

To take the simplest example, suppose that, after randomisation, the treatment group is different in significant ways from the control group. This was the case in a 1995 trial showing that a clot busting drug was effective for ischaemic stroke.2 The treatment group were 3% …