Comparison of Parametric and Non-Parametric Survival Methods Using Simulated Clinical Data

John W. Gamel; Robert L. Vogel

Comparison of Parametric and Non-Parametric Survival Methods Using Simulated Clinical Data

John W. Gamel, Robert L. Vogel

Emeriti, Retired, & In Memoriam (JPHCOPH)

Research output: Contribution to journal › Article › peer-review

17 Scopus citations

Abstract

<div class="line" id="line-5"> We derived three parametric survival models (the log&hyphen;normal, log logit, and Weibull) from the clinical data of chemotherapy trials for stage II breast cancer. We then used these models to generate simulated survival data, which we analysed using both parametric (log&hyphen;normal) and non&hyphen;parametric (logrank, Gray–Tsiatis and Laska–Meisner) methods. With limited follow&hyphen;up (5 years), the non&hyphen;parametric tests had greater power than the log&hyphen;normal model. This advantage diminished, however, with extended follow&hyphen;up (15 years). Furthermore, only the log&hyphen;normal model could distinguish reliably a survival advantage due to an increase in cured fraction from an advantage due to an increase in time to failure.</div>

Original language	American English
Journal	Statistics in Medicine
Volume	16
State	Published - Jul 1 1997

Keywords

Comparison
Non-parametric
Parametric
Simulated clinical data
Survival methods

DC Disciplines

Public Health

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Cite this

@article{d2008228b3e34ac4bbefb5fbba30ae12,

title = "Comparison of Parametric and Non-Parametric Survival Methods Using Simulated Clinical Data",

abstract = " We derived three parametric survival models (the log&hyphen;normal, log logit, and Weibull) from the clinical data of chemotherapy trials for stage II breast cancer. We then used these models to generate simulated survival data, which we analysed using both parametric (log&hyphen;normal) and non&hyphen;parametric (logrank, Gray–Tsiatis and Laska–Meisner) methods. With limited follow&hyphen;up (5 years), the non&hyphen;parametric tests had greater power than the log&hyphen;normal model. This advantage diminished, however, with extended follow&hyphen;up (15 years). Furthermore, only the log&hyphen;normal model could distinguish reliably a survival advantage due to an increase in cured fraction from an advantage due to an increase in time to failure.",

keywords = "Comparison, Non-parametric, Parametric, Simulated clinical data, Survival methods",

author = "Gamel, {John W.} and Vogel, {Robert L.}",

note = "Enter your email address below. If your address has been previously registered, you will receive an email with instructions on how to reset your password. If you don't receive an email, you should register as a new user",

year = "1997",

month = jul,

day = "1",

language = "American English",

volume = "16",

journal = "Statistics in Medicine",

issn = "0277-6715",

publisher = "John Wiley and Sons Ltd",

}

TY - JOUR

T1 - Comparison of Parametric and Non-Parametric Survival Methods Using Simulated Clinical Data

AU - Gamel, John W.

AU - Vogel, Robert L.

N1 - Enter your email address below. If your address has been previously registered, you will receive an email with instructions on how to reset your password. If you don't receive an email, you should register as a new user

PY - 1997/7/1

Y1 - 1997/7/1

N2 - We derived three parametric survival models (the log&hyphen;normal, log logit, and Weibull) from the clinical data of chemotherapy trials for stage II breast cancer. We then used these models to generate simulated survival data, which we analysed using both parametric (log&hyphen;normal) and non&hyphen;parametric (logrank, Gray–Tsiatis and Laska–Meisner) methods. With limited follow&hyphen;up (5 years), the non&hyphen;parametric tests had greater power than the log&hyphen;normal model. This advantage diminished, however, with extended follow&hyphen;up (15 years). Furthermore, only the log&hyphen;normal model could distinguish reliably a survival advantage due to an increase in cured fraction from an advantage due to an increase in time to failure.

AB - We derived three parametric survival models (the log&hyphen;normal, log logit, and Weibull) from the clinical data of chemotherapy trials for stage II breast cancer. We then used these models to generate simulated survival data, which we analysed using both parametric (log&hyphen;normal) and non&hyphen;parametric (logrank, Gray–Tsiatis and Laska–Meisner) methods. With limited follow&hyphen;up (5 years), the non&hyphen;parametric tests had greater power than the log&hyphen;normal model. This advantage diminished, however, with extended follow&hyphen;up (15 years). Furthermore, only the log&hyphen;normal model could distinguish reliably a survival advantage due to an increase in cured fraction from an advantage due to an increase in time to failure.

KW - Comparison

KW - Non-parametric

KW - Parametric

KW - Simulated clinical data

KW - Survival methods

UR - https://doi.org/10.1002/(SICI)1097-0258(19970730)16:14%3C1629::AID-SIM594%3E3.0.CO;2-C

M3 - Article

SN - 0277-6715

VL - 16

JO - Statistics in Medicine

JF - Statistics in Medicine

ER -

Comparison of Parametric and Non-Parametric Survival Methods Using Simulated Clinical Data

Abstract

Keywords

DC Disciplines

UN SDGs

Other files and links

Fingerprint

Cite this