Statistics and Its Interface

Volume 9 (2016)

Number 3

Multistage nonparametric tests for treatment comparisons in clinical trials with multiple primary endpoints

Pages: 343 – 354



Peng Huang (Department of Oncology and Department of Biostatistics, Johns Hopkins University, Baltimore, Maryland, U.S.A.)

Ming T. Tan (Department of Biostatistics, Bioinformatics and Biomathematics, Georgetown University, Washington, D.C., U.S.A.)


Many clinical trials, e.g., neurodegenerative disease trials, are conducted to test whether a new treatment could slow or modify disease progression. Multiple primary endpoints are often used since it is difficult to find a single clinical endpoint that summarizes the treatment effect, e.g., the neuroprotective effect. There are three major challenges in the design and analysis of such trials: (1) the presence of nuisance effect regardless whether the desired neuroprotective effect exists; (2) primary endpoints are of mixed type; (3) the need for interim analysis stopping rule for multiple primary endpoints. We propose a simple nonparametric multistage adaptive (group sequential) test to overcome these difficulties. Statistically, this test is another solution to the multivariate nonparametric Behrens-Fisher problem. We provide both large and small sample properties of the proposed test. The methodology is illustrated using data from two randomized clinical trials.


rank-based test, Behrens-Fisher problem, adaptive group sequential test, Brownian motion

2010 Mathematics Subject Classification

Primary 62G10. Secondary 62H15, 62L05.

Full Text (PDF format)

Published 27 January 2016