*Geek Box: Crossover Design
In a traditional comparative trial, a parallel design is often employed, where the intervention group receives a treatment, and the control group is either a placebo or other standard treatment [like habitual diet]; both arms run through the trial at the same time. In a crossover design, all participants receive both treatments sequentially, separated by a washout period, i.e., Treatment A – Washout – Treatment B. Randomisation means that the order of treatments is random, i.e., some participants start with Treatment B then switch to Treatment A, and vice versa. Generally, crossover studies are appropriate where the effects of the treatments will be short-term, such that a washout period may have the effect of returning to pre-treatment baseline. However, this is not always the case, and there may be a legacy, carryover effect of treatments, meaning it is important to statistically analyse any potential interaction of treatment order. Crossover designs are, however, quite useful for nutrition science given the inherent issue of a lack of true placebo for food, which means we’re left comparing differences in the same dietary variables. Crossover designs can be helpful in this respect, to compare Diet A vs. Diet B, and manipulate specific variables between the two treatments.