[ABE-L] Seminário Conjunto UFSCar/ICMC - 27/03/2015 - 14h00 - na UFSCar

[Rafael Izbicki]

*Seminário Conjunto UFSCar/ICMC – 27/03/2015 - 14h00*

*TÍTULO:* A Non-inferiority Test for Survival Response Assuring the Control
of Type I Error

*PALESTRANTE: *Juliana Cobre - ICMC/USP

*LOCAL: *Sala de seminários – DEs-UFSCar
*RESUMO: *The bio-pharmaceutical industry has been developing new
treatments that, compared to a standard treatment of proven efficacy, have
characteristics such as greater quality of life, lower cost, easier
administration, decreased side effects. Even  the new treatment being worse
in terms of the survival outcome it can be approved if it has such
advantages. In this cases a non-inferiority hypothesis test, instead of a
superiority or an equivalent test, must be conducted. Often the
non-inferiority trials for survival response assume that the data follow a
proportional hazard model (PHM). Nevertheless we prove here that if this
assumption is not true the usual log-rank based non-inferiority tests
inflate the type I error rate, rather when the sample size goes to
infinity, the limit of the type I error rate is equal to 1 at the margin of
the null space. To circumvent this serious problem, we propose a
proportional odds survival model (POSM) based non-inferiority hypothesis
test. The main advantage of our test is that it controls the type I error
rate when the true model is both PHM and POSM. The sample sizes
determination to reach a desired power complete our proposal.


--
Rafael Izbicki
Assistant Professor
Department of Statistics
Federal University of São Carlos (UFSCar)
http://www.rizbicki.ufscar.br/
-------------- Próxima Parte ----------
Um anexo em HTML foi limpo...
URL: <https://lists.ime.usp.br/archives/abe/attachments/20150323/0bdd2775/attachment.html>