Question normal

Since matching was not done in this study, the confounding variables could have been accounted for by crossover design. Perhaps it could also have reduced the required sample size. Are there any shortcomings of a crossover design [ other than longer study period] that led to the choice of parallel groups?