×

Exact two-sample permutation tests based on the randomized play-the- winner rule. (English) Zbl 0657.62086

Comparing two treatments in a clinical trial where patients are supposed to enter sequentially is not only a purely mathematical problem. Optimum designs for such trials assign half the patients to the inferior treatment on average and may therefore be less desirable ethically. The latter is taken into account by so-called play-the-winner rules: a success on a particular treatment generates a future trial on the same treatment with a new patient; a failure on a treatment generates a future trial on the alternative treatment. This rule can be modeled by an urn scheme. Though this rule tends to put more patients on the better treatment it is too deterministic and may introduce bias to the trial.
The author [Ann. Stat. 7, 1979, 291-296 (1979; Zbl 0399.62083)] introduced the randomized play-the-winner rule which is a modification of the rule described above. In the present paper, exact permutation tests based on the randomized play-the-winner rule are constructed using an efficient network algorithm. The test procedure is illustrated with a prospective controlled randomized study of the use of extracorporal membrane oxygenation to treat newborns with respiratory failure.
Reviewer: D.Rasch

MSC:

62L05 Sequential statistical design
62P10 Applications of statistics to biology and medical sciences; meta analysis
62F03 Parametric hypothesis testing

Citations:

Zbl 0399.62083
Full Text: DOI