In pediatric drug development, large amounts of adult data are often available before the start of a pediatric study. It is believed that borrowing this information will improve the efficiency. However, when adult information is not sufficiently similar to that of pediatrics, incorporating adult data will introduce bias and consequently result in efficiency loss. A Bayesian alternative-namely, commensurate prior approach where the level of information borrowing is based on the concordance of adult and pediatric data-was investigated. Simulation results indicate that the commensurate prior approach, in general, provides a balanced and robust trade-off between bias and efficiency gain. The benefit of this approach was quantified in terms of sample size savings, and recommended sample sizes are provided.
Keywords: Bayesian analysis; borrowing; commensurate prior; historical data; pediatrics; sample size determination.
© 2019, The American College of Clinical Pharmacology.