Non-inferiority test based on transformations for non-normal distributions

Santu Ghosh, Arpita Chatterjee, Samiran Ghosh

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Non-inferiority trials are becoming very popular for comparative effectiveness research. These trials are required to show that the effect of an experimental treatment is not worse than that of a reference treatment by more than a specified margin. Hence non-inferiority trials are of great importance, when superiority cannot be claimed. A three-arm non-inferiority trial consists of a placebo, a reference treatment, and an experimental treatment is considered. However unlike the traditional choices, it is assumed that the distributions of the end points corresponding to these treatments are unknown and suggested test procedures for a three-arm non-inferiority trial based on monotone transformations in conjunction with a normal approximation. The resulting test procedures are flexible and robust. Theoretical properties of the proposed methods are also investigated. The performance of the suggested test procedures is compared to their counterparts using simulations. In terms of type I error and power, the proposed methods perform better than their counterparts in most cases. The usefulness of the proposed methods is further illustrated through an example.

Original languageEnglish (US)
Pages (from-to)73-87
Number of pages15
JournalComputational Statistics and Data Analysis
Volume113
DOIs
StatePublished - Sep 2017

Fingerprint

Non-inferiority
Non-normal Distribution
Normal Approximation
Type I error
End point
Margin
Monotone
Unknown
Simulation

Keywords

  • Edgeworth expansion
  • Level error
  • Non-inferiority
  • Rank test
  • Three-arm trial

ASJC Scopus subject areas

  • Statistics and Probability
  • Computational Mathematics
  • Computational Theory and Mathematics
  • Applied Mathematics

Cite this

Non-inferiority test based on transformations for non-normal distributions. / Ghosh, Santu; Chatterjee, Arpita; Ghosh, Samiran.

In: Computational Statistics and Data Analysis, Vol. 113, 09.2017, p. 73-87.

Research output: Contribution to journalArticle

@article{404c3d59adcb449885169093e8a9a030,
title = "Non-inferiority test based on transformations for non-normal distributions",
abstract = "Non-inferiority trials are becoming very popular for comparative effectiveness research. These trials are required to show that the effect of an experimental treatment is not worse than that of a reference treatment by more than a specified margin. Hence non-inferiority trials are of great importance, when superiority cannot be claimed. A three-arm non-inferiority trial consists of a placebo, a reference treatment, and an experimental treatment is considered. However unlike the traditional choices, it is assumed that the distributions of the end points corresponding to these treatments are unknown and suggested test procedures for a three-arm non-inferiority trial based on monotone transformations in conjunction with a normal approximation. The resulting test procedures are flexible and robust. Theoretical properties of the proposed methods are also investigated. The performance of the suggested test procedures is compared to their counterparts using simulations. In terms of type I error and power, the proposed methods perform better than their counterparts in most cases. The usefulness of the proposed methods is further illustrated through an example.",
keywords = "Edgeworth expansion, Level error, Non-inferiority, Rank test, Three-arm trial",
author = "Santu Ghosh and Arpita Chatterjee and Samiran Ghosh",
year = "2017",
month = "9",
doi = "10.1016/j.csda.2016.10.004",
language = "English (US)",
volume = "113",
pages = "73--87",
journal = "Computational Statistics and Data Analysis",
issn = "0167-9473",
publisher = "Elsevier",

}

TY - JOUR

T1 - Non-inferiority test based on transformations for non-normal distributions

AU - Ghosh, Santu

AU - Chatterjee, Arpita

AU - Ghosh, Samiran

PY - 2017/9

Y1 - 2017/9

N2 - Non-inferiority trials are becoming very popular for comparative effectiveness research. These trials are required to show that the effect of an experimental treatment is not worse than that of a reference treatment by more than a specified margin. Hence non-inferiority trials are of great importance, when superiority cannot be claimed. A three-arm non-inferiority trial consists of a placebo, a reference treatment, and an experimental treatment is considered. However unlike the traditional choices, it is assumed that the distributions of the end points corresponding to these treatments are unknown and suggested test procedures for a three-arm non-inferiority trial based on monotone transformations in conjunction with a normal approximation. The resulting test procedures are flexible and robust. Theoretical properties of the proposed methods are also investigated. The performance of the suggested test procedures is compared to their counterparts using simulations. In terms of type I error and power, the proposed methods perform better than their counterparts in most cases. The usefulness of the proposed methods is further illustrated through an example.

AB - Non-inferiority trials are becoming very popular for comparative effectiveness research. These trials are required to show that the effect of an experimental treatment is not worse than that of a reference treatment by more than a specified margin. Hence non-inferiority trials are of great importance, when superiority cannot be claimed. A three-arm non-inferiority trial consists of a placebo, a reference treatment, and an experimental treatment is considered. However unlike the traditional choices, it is assumed that the distributions of the end points corresponding to these treatments are unknown and suggested test procedures for a three-arm non-inferiority trial based on monotone transformations in conjunction with a normal approximation. The resulting test procedures are flexible and robust. Theoretical properties of the proposed methods are also investigated. The performance of the suggested test procedures is compared to their counterparts using simulations. In terms of type I error and power, the proposed methods perform better than their counterparts in most cases. The usefulness of the proposed methods is further illustrated through an example.

KW - Edgeworth expansion

KW - Level error

KW - Non-inferiority

KW - Rank test

KW - Three-arm trial

UR - http://www.scopus.com/inward/record.url?scp=85006010318&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85006010318&partnerID=8YFLogxK

U2 - 10.1016/j.csda.2016.10.004

DO - 10.1016/j.csda.2016.10.004

M3 - Article

AN - SCOPUS:85006010318

VL - 113

SP - 73

EP - 87

JO - Computational Statistics and Data Analysis

JF - Computational Statistics and Data Analysis

SN - 0167-9473

ER -