Designing a randomized clinical trial (RCT) that incorporates flexible approaches to sample size calculation, data monitoring, early stopping, multiple endpoints, and changing sample size during the trial all while still preserving the stated a-level is computational difficult. Therefore, a computerized approach to the design of an RCT has important commercial application to pharmaceutical companies, both domestic and foreign, and other organizations that perform clinical trials. Phase I of this SBIR studied a new class of designs for an RCT. The new design designates a portion of the main clinical trial as an internal pilot (IP) study. The sample size is recalculated o the basis of data from the IP. Phase II will continue the work of Phase I in investigating the operating characteristics of IPs for the normal, the binomial, and the log-rank statistics. It will develop a computer program that implements IP studies incorporating the Lakatos method of sample size calculation, the Lan-DeMets a-spending approach to data monitoring, stochastic curtailing, and Bonferroni correction for multiple endpoints. The research component of Phase II will focus on statistical methods for combining the approaches while still maintaining control over alpha and power.
| Proschan, M A; Wittes, J (2000) An improved double sampling procedure based on the variance. Biometrics 56:1183-7 |
| Zucker, D M; Wittes, J T; Schabenberger, O et al. (1999) Internal pilot studies II: comparison of various procedures. Stat Med 18:3493-509 |
