Sample size and power calculations based on generalized linear mixed models with correlated binary outcomes

Authors:
Qianyu Dang;Sati Mazumdar;Patricia R. Houck
Affiliations:
Department of Medicine, University of Pittsburgh, Pittsburgh, PA 15213, United States and Department of Biostatistics, University of Pittsburgh, Pittsburgh, PA 15213, United States;Department of Biostatistics, University of Pittsburgh, Pittsburgh, PA 15213, United States and Department of Psychiatry, University of Pittsburgh, Pittsburgh, PA 15213, United States;Department of Psychiatry, University of Pittsburgh, Pittsburgh, PA 15213, United States
Venue:
Computer Methods and Programs in Biomedicine
Year:
2008

Citing 1
Cited 0

Power calculations for generalized linear models in observational longitudinal studies: A simulation approach in SAS

Computer Methods and Programs in Biomedicine

Quantified Score

Hi-index	0.00

Visualization

Abstract

The generalized linear mixed model (GLIMMIX) provides a powerful technique to model correlated outcomes with different types of distributions. The model can now be easily implemented with SAS PROC GLIMMIX in version 9.1. For binary outcomes, linearization methods of penalized quasi-likelihood (PQL) or marginal quasi-likelihood (MQL) provide relatively accurate variance estimates for fixed effects. Using GLIMMIX based on these linearization methods, we derived formulas for power and sample size calculations for longitudinal designs with attrition over time. We found that the power and sample size estimates depend on the within-subject correlation and the size of random effects. In this article, we present tables of minimum sample sizes commonly used to test hypotheses for longitudinal studies. A simulation study was used to compare the results. We also provide a Web link to the SAS macro that we developed to compute power and sample sizes for correlated binary outcomes.