Abstract:
|
An important measure of association, the relative risk is used to compare the results of two treatments or exposures. Standard approaches to estimating relative risk available in common software packages may produce biased inferences when applied to correlated binary data collected from longitudinal or clustered studies. In recent years, several methods for estimating the risk ratio for correlated binary data have been published, some of which maintain a well-controlled coverage probability but do not maintain an appropriate interval width or the interval location to measure the balance between distal and mesial noncoverage probabilities accurately or, vice versa. We propose procedures for constructing a confidence interval for risk ratio that directly extends recently recommended methods for correlated binary data by building on the concepts of the design effect and effective sample sizes typically used in representative sample surveys. In order to investigate the performance of these proposed methods, we conduct an extensive simulation study. We illustrate the usefulness of our proposed methods with an example from real-life application
|