Paired binary data arise in studies with two measures on
the same subject in before-and-after trials, two-period crossover
trials, and matched case-control studies. For a binary
outcome (eg, yes/no), a 2 2 table with the same row and
column categories summarizes the data. Although
repeated-measures designs can consist ofmultiple measures
on the same subject, the focus of this article is on those designs with only two measures per subject, with no covariates.
A key feature of paired binary data is that these two
observations from the same individual tend to be positively
correlated, which must be accounted for in the design and
analysis of the study [3,4]. With the increasing use of marginal
models, such as generalized estimating equation
(GEE) models, and introduction of the procedure GLIMMIX
for conditional models by SAS version 9.1, researchers
may be unclear on how to estimate sample size
for studies with a paired binary primary outcome.