Application of Sample Survey Methods for Modeling Ratios to Incidence Data
Lisa M. LaVange,
Gary G. Koch, Lynette L. Keyes, and P. A. Margolis
1994, Statistics
in Medicine
13, 343-355
Keywords:
Incidence density analysis, weighted least squares, adverse event associations
Abstract
We describe
ratio estimation methods for analyzing incidence data from follow-up
studies. Commonly used in survey data analysis, these ratio methods
require minimal distributional assumptions and accurately account for
random variability in the at-risk periods and correlations among
repeated events. The methods are easy to understand, readily
available via commercial software, and provide flexibility for a
variety of analytical settings. We suggest that ratio methods may be
useful for epidemiological and clinical studies in which quantities
such as incidence of illness events or side effects of drug treatment
are the focus. The basic strategy consists of a two-step process in
which we first estimate subgroup specific incidence densities and
their covariance matrix via a first order Taylor series
approximation. We then fit log-linear models to the estimated ratios
in order to assess covariate effects. The ability to produce direct
estimates of adjusted incidence density ratios is an important
advantage of this approach. We provide illustrative analyses of
incidence data using ratio methods as well as survey logistic
regression methods and two applications of generalized estimating
equation methodology, repeated logistic and Poisson regression
models, for comparison.