Hua He, of the Department of Epidemiology in the School of Public Health and Tropical Medicine at Tulane University, will present:
“Statistical Analysis of Censored Data Due to Detection Limits”
Abstract: Measures of substance concentration in urine, serum or other biological matrices often have an assay limit of detection. When concentration levels fall below the limit, the exact measures cannot be obtained, and thus are left censored. Common practices for handling censored data, such as replacing the censored data with a constant value or deleting the censored data, often yield biased and/or inefficient estimates. The problems become more challenging when the censored data come from heterogenous populations consisting of exposure and non-exposure subpopulations. When the censored data come from non-exposed subjects, their measures are not observed and thus form a latent class resulting from a different censoring mechanism than that involved with exposed subjects. In this talk, I will discuss statistical analysis of the censored data when treated as either outcomes or predictors, such as the Tobit model, a mixture Tobit model for censored outcomes, and joint modeling for censored predictors. I will also talk about several tests for testing latent class in the censored data as well as some ongoing and future work. Some results based on simulation studies and real data examples will be presented to illustrate the methods, and some guidelines for analyzing censored data due to detection limit will be provided as well.
All are welcome.