Crossover Design

# Paralled Study Design: each subject is randomized to one and only one treatment

# Crossover Design: each subject receives more than one treatment in a specified sequene

# Crossover Trials assumes that patients usually have a chronically stable condition that will not vary between when they are taking the 1st and 2nd treatments.

# p*q crossover design: p sequences of treatments administered over q different dosing periods.

### Advantages of crossover trials over parallel studies

# Treatment differences can be based on within-subject comparisons instead of between-subject comparisons (less variability is expected).

# require less sample size
N(cross over) = (1-r)*N(paralle) / 2 , where r is correlation coefficient

# a crossover design provides the least-biased estimates for the difference between treatments assuming that the response of subjects to treatment is consistent

### Main limitations of crossver trials

# crossover trials pose greater inconvenience to the subjects
# censored observations due to subject withdrawal have a higher impact
# subjects should be in a comparable condition at the start of each treatment period; it is more appropriate for chronic disease that have a stable set of symptoms, while it is not suitable either for acute conditions, for primary outcomes that are permanent or for terminal events (pregnancy or death).
# it requires fewer patients, but not for Phase III study in which needs large sample size anyway.
# carryover effect
# period effect (treatment by period interaction)