In this section, we've seen how performance bias, ascertainment bias and attrition bias could affect the validity of an RCT. Performance bias means any difference in the way the groups were treated (apart from the experimental intervention). Ascertainment bias means any difference in the way the outcomes were assessed between the groups.

The most effective countermeasure against these sorts of bias is "masking" (or blinding). You should check whether all of the people who could have influenced the outcome of the trial were blinded to treatment. If it was not possible to mask all of the participants, you should consider whether the researchers took appropriate steps to make sure that the other participants were masked.

Attrition bias may arise where patients are lost to follow-up or crossover between treatment groups. It's essential that a study follow up a high enough proportion of the original sample to be able to make valid claims about effectiveness. It's also essential that a valid study report its results as an "intention-to-treat" analysis.