In a crossover trial, both groups are exposed to the intervention and to the placebo at different times, or both groups are exposed to an intervention but in a different order.
Deeper dive
The purpose of using randomization to divide study participants into groups in a randomized controlled trial is to avoid bias in how the participants are assigned to an intervention or placebo.
But even with randomization, groups can end up being different enough to potentially influence the reliability of the results. For example, one group could end up with more comorbidities (e.g., a higher proportion of participants with Type 2 diabetes) or a significantly higher proportion of a certain demographic (e.g., more low-income patients). Or, a trial could be so small — perhaps only one or two dozen people — that even seemingly minor differences between the groups could affect the results.
In other trials, it may not be that the participants in different arms of a study are different but that receiving interventions in one order has different results than if the treatments had been administered in a different order. In any of these cases, researchers might use a crossover trial. In one type of crossover trial, one group (arm) receives the intervention while the other receives the placebo for the first half of the trial. Then, usually after a brief washout period, the groups switch. This study design is frequently used in diet studies, where individual differences can profoundly affect the results, and in trials where the passing of time itself might influence the results.
The other type of crossover study, in which interventions are administered to all study groups but in a different order, is common in cancer studies. One example is patients receiving radiation before chemo vs. chemo before radiation, which could have dramatically different outcomes. In these trials, there may not be a placebo group or traditional “control” group at all. Both the groups basically receive the same treatment, but the crossover design allows researchers to see if the sequence of interventions makes a difference.