Nonparametric estimation of duration of accrual and total study length for clinical trials.

Nonparametric estimates for all possible values of accrual duration and total study length required to achieve a specified power and level of significance are given assuming a proportional hazards model comparing two treatment groups. Based on data from a previous representative clinical trial, the nonparametric maximum likelihood estimate of total study length and its asymptotic distribution are derived as functions of accrual duration. An example using this technique is given for designing a clinical trial comparing two adjuvant chemotherapy regimens for large bowel cancer.