Abstract:
|
In clinical trials, handling missing data correctly is critical to drawing unbiased inferences on treatment effects. In our research, we focused on longitudinal count data with dropout in clinical trials, and evaluated various imputation models in MI (multiple imputation) for incomplete longitudinal count data, as such research is still limited. We developed programming code for some imputation methods, which is based on full parametric models (e.g. Poisson regression and negative binomial regression model) and nonparametric methods (e.g. nearest neighbor matching and propensity score stratification matching) for monotone missing patterns. Specifically, sensitivity of parametric methods to deviations in distributional assumptions (e.g. overdispersion) was contrasted with that of nonparametric methods. An example motivated by an actual clinical trial was used in the evaluations. Simulation studies were performed under various settings including missing mechanisms and missing fractions. The details of results will be shown in the presentation. This work was supported by JSPS KAKENHI Grant Number 25240005.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.