Review Question - QID 103901

The utility of a crossover study lies in the fact that variation between participants is minimized, thereby increasing the precision and accuracy of the estimate of the effect size. Additionally, equipoise is established by ensuring all participants have access to the therapy in question.

When conducing controlled trials, patients are randomized to ensure that unmeasured confounders between the intervention arm and the placebo arm are roughly the same. Crossover designs improve upon this further by ensuring unmeasured confounders are identical, because all participants ultimately provide data under both circumstances. Thus, a single participant's co-morbidity cannot affect the results, because it would affect both arms equally as he/she experiences both arms of the study. In simplified terms, the superiority of crossover designs can be thought to stem from the fact that participants serve as their own controls.

Rosner describes the major concern of crossover designs: carry-over effect. The major limitation of crossover designs is the fact that patients originally assigned to the intervention arm may still derive residual benefit from the drug even while they have started taking placebo. This concern is obviously greater for drugs with long durations of action and half-lives. The solution to this problem is the washout period. Ideally, the washout period should be sufficiently long to ensure that patients have completely cleared any of the effect of the intervention before their placebo period begins.

Gordis describes how crossover designs may enter into the analysis of studies not originally intended to include it. In controlled trials, patients are randomized to either receive an intervention or not. However, the vagaries of life sometimes prevent participants assigned to the intervention from actually receiving it, e.g. they stop taking the drug. In some cases, the participants who are assigned to placebo end up receiving the intervention under investigation through medical care provided outside of the study. In both cases, the participants start in one arm of the study but end up in a different arm. A crossover analysis can account for their switching.

Illustration A is a general schematic for a crossover study in which a drug and a placebo are given at alternate times in two separate study arms.

Incorrect Answers:
Answer 1: Crossover studies expose all participants to the intervention and establish equipoise, but this is not the only correct answer here.
Answer 2: Crossover studies minimize variations between participants, but this is not the only correct answer here.
Answer 3: The purpose of the washout period is to decrease carry-over effects.
Answer 5: Incorrect because of Answer 3.