Power analysis for clustered non-continuous responses in multicenter trials

T. Chen, K. Knox, J. Arora, W. Tang, J. Kowalski, X. M. Tu

Research output: Contribution to journalArticle

Abstract

Power analysis for multi-center randomized control trials is quite difficult to perform for non-continuous responses when site differences are modeled by random effects using the generalized linear mixed-effects model (GLMM). First, it is not possible to construct power functions analytically, because of the extreme complexity of the sampling distribution of parameter estimates. Second, Monte Carlo (MC) simulation, a popular option for estimating power for complex models, does not work within the current context because of a lack of methods and software packages that would provide reliable estimates for fitting such GLMMs. For example, even statistical packages from software giants like SAS do not provide reliable estimates at the time of writing. Another major limitation of MC simulation is the lengthy running time, especially for complex models such as GLMM, especially when estimating power for multiple scenarios of interest. We present a new approach to address such limitations. The proposed approach defines a marginal model to approximate the GLMM and estimates power without relying on MC simulation. The approach is illustrated with both real and simulated data, with the simulation study demonstrating good performance of the method.

Original languageEnglish (US)
Pages (from-to)979-995
Number of pages17
JournalJournal of Applied Statistics
Volume43
Issue number6
DOIs
StatePublished - Apr 25 2016
Externally publishedYes

Fingerprint

Power Analysis
Linear Mixed Effects Model
Monte Carlo Simulation
Estimate
Statistical package
Marginal Model
Generalized Linear Mixed Model
Sampling Distribution
Power Function
Random Effects
Software Package
Extremes
Simulation Study
Scenarios
Software
Model
Monte Carlo simulation

Keywords

  • GEE
  • GLIMMIX
  • intraclass correlation
  • marginal models
  • NLMIXED

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Power analysis for clustered non-continuous responses in multicenter trials. / Chen, T.; Knox, K.; Arora, J.; Tang, W.; Kowalski, J.; Tu, X. M.

In: Journal of Applied Statistics, Vol. 43, No. 6, 25.04.2016, p. 979-995.

Research output: Contribution to journalArticle

Chen, T. ; Knox, K. ; Arora, J. ; Tang, W. ; Kowalski, J. ; Tu, X. M. / Power analysis for clustered non-continuous responses in multicenter trials. In: Journal of Applied Statistics. 2016 ; Vol. 43, No. 6. pp. 979-995.
@article{302ccda9d431496ca6b60fec0580ff54,
title = "Power analysis for clustered non-continuous responses in multicenter trials",
abstract = "Power analysis for multi-center randomized control trials is quite difficult to perform for non-continuous responses when site differences are modeled by random effects using the generalized linear mixed-effects model (GLMM). First, it is not possible to construct power functions analytically, because of the extreme complexity of the sampling distribution of parameter estimates. Second, Monte Carlo (MC) simulation, a popular option for estimating power for complex models, does not work within the current context because of a lack of methods and software packages that would provide reliable estimates for fitting such GLMMs. For example, even statistical packages from software giants like SAS do not provide reliable estimates at the time of writing. Another major limitation of MC simulation is the lengthy running time, especially for complex models such as GLMM, especially when estimating power for multiple scenarios of interest. We present a new approach to address such limitations. The proposed approach defines a marginal model to approximate the GLMM and estimates power without relying on MC simulation. The approach is illustrated with both real and simulated data, with the simulation study demonstrating good performance of the method.",
keywords = "GEE, GLIMMIX, intraclass correlation, marginal models, NLMIXED",
author = "T. Chen and K. Knox and J. Arora and W. Tang and J. Kowalski and Tu, {X. M.}",
year = "2016",
month = "4",
day = "25",
doi = "10.1080/02664763.2015.1089218",
language = "English (US)",
volume = "43",
pages = "979--995",
journal = "Journal of Applied Statistics",
issn = "0266-4763",
publisher = "Routledge",
number = "6",

}

TY - JOUR

T1 - Power analysis for clustered non-continuous responses in multicenter trials

AU - Chen, T.

AU - Knox, K.

AU - Arora, J.

AU - Tang, W.

AU - Kowalski, J.

AU - Tu, X. M.

PY - 2016/4/25

Y1 - 2016/4/25

N2 - Power analysis for multi-center randomized control trials is quite difficult to perform for non-continuous responses when site differences are modeled by random effects using the generalized linear mixed-effects model (GLMM). First, it is not possible to construct power functions analytically, because of the extreme complexity of the sampling distribution of parameter estimates. Second, Monte Carlo (MC) simulation, a popular option for estimating power for complex models, does not work within the current context because of a lack of methods and software packages that would provide reliable estimates for fitting such GLMMs. For example, even statistical packages from software giants like SAS do not provide reliable estimates at the time of writing. Another major limitation of MC simulation is the lengthy running time, especially for complex models such as GLMM, especially when estimating power for multiple scenarios of interest. We present a new approach to address such limitations. The proposed approach defines a marginal model to approximate the GLMM and estimates power without relying on MC simulation. The approach is illustrated with both real and simulated data, with the simulation study demonstrating good performance of the method.

AB - Power analysis for multi-center randomized control trials is quite difficult to perform for non-continuous responses when site differences are modeled by random effects using the generalized linear mixed-effects model (GLMM). First, it is not possible to construct power functions analytically, because of the extreme complexity of the sampling distribution of parameter estimates. Second, Monte Carlo (MC) simulation, a popular option for estimating power for complex models, does not work within the current context because of a lack of methods and software packages that would provide reliable estimates for fitting such GLMMs. For example, even statistical packages from software giants like SAS do not provide reliable estimates at the time of writing. Another major limitation of MC simulation is the lengthy running time, especially for complex models such as GLMM, especially when estimating power for multiple scenarios of interest. We present a new approach to address such limitations. The proposed approach defines a marginal model to approximate the GLMM and estimates power without relying on MC simulation. The approach is illustrated with both real and simulated data, with the simulation study demonstrating good performance of the method.

KW - GEE

KW - GLIMMIX

KW - intraclass correlation

KW - marginal models

KW - NLMIXED

UR - http://www.scopus.com/inward/record.url?scp=84958929418&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84958929418&partnerID=8YFLogxK

U2 - 10.1080/02664763.2015.1089218

DO - 10.1080/02664763.2015.1089218

M3 - Article

AN - SCOPUS:84958929418

VL - 43

SP - 979

EP - 995

JO - Journal of Applied Statistics

JF - Journal of Applied Statistics

SN - 0266-4763

IS - 6

ER -