Back to Annual Meeting Page		American Public Health Association 133rd Annual Meeting & Exposition December 10-14, 2005 Philadelphia, PA

Abstract #122211

Hormuzd Katki, Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Cancer Institute, NIH, DHHS, 6120 Executive Blvd. Room 8044 MSC 7244, Rockville, MD 20852-4910, 301-594-7818, katkih@mail.nih.gov and Steven D. Mark, MD, PhD, Department of Preventive Medicine and Biometrics, SOM, University of Colorado Health Sciences Center, 4200 E. 9th Ave., B-119, Denver, CO 80262.

In studies nested within cohorts, the censored survival outcomes and easily obtainable covariates are observed on all members, but certain expensive or difficult to obtain covariates are observed only on a subsample. Examples include two-stage, case-cohort, and nested case-control designs. Our software is based on Mark and Katki (JASA, in press) and extends and unifies current analysis options in several ways. Our software fits Cox s and Kaplan-Meier curves to estimate absolute and population attributable risks for multiple exposures, standardized for confounders. We allow the subsampling scheme to depend on any cohort information: in particular you can subsample cases and frequency-match or counter-match. We improve efficiency by using a more efficient weighting scheme and by using (if available) surrogates for exposure. The estimators are valid as long as the subsampling scheme is known. Our software is freely available in the R language. We demonstrate our software on a cohort study of esophageal cancer and zinc, where study constraints limited zinc measurement to only 25% of the cohort.

Learning Objectives: The viewer will be able to

• Understand the general framework of case-cohort and related studies
• Discern under what circumstances this software will be useful

Keywords: Statistics,

Presenting author's disclosure statement:

Not Answered