Simultaneous Confidence Intervals for Proportion Difference from Correlated Bilateral Data in Ophthalmologic Studies

• Activity Number: 136
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Zhengyu Yang* and Xiaobin Liu and Song Liu and Changxing Ma
• Companies: SUNY Buffalo and SUNY Buffalo and Roswell Park Cancer Institute and SUNY Buffalo
• Keywords: Confidence interval ; Bilateral data ; Difference in proportions ; Wald method ; Profile likelihood method ; Adjustment for multiplicity

Abstract

Bilateral correlated data often arise in ophthalmological studies, where each subject may contribute results of two eyes to the analysis. Ignoring the correlation between eyes can yield biased inference. When comparing several treatments to a general control, it is often of interest to construct simultaneous confidence intervals for difference of proportions. We propose four simultaneous confidence intervals with multiplicity adjustment, taking the intraclass correlation into account. The coverage probabilities and widths of the resulting methods are compared in a Monte Carlo simulation study. A Real working example is presented to illustrate the methodology.