ST07 Automated Summarization of Risk Assessment of Discrete and Continuous Exposures in Epidemiological Studies using SAS(R) ODS     Contributed

Stuart Long Westat ,Rebecca Darden
Westat
Abstract: In epidemiological research, we frequently deal with case-control studies where the researcher wishes to investigate associations between the outcome and numerous potential exposures added to a predetermined set of base model covariates. Exposures may be c ontinuous, dichotomous, or multi-level. Our macro runs a PROC LOGISTIC model with adjustments for predetermined base model covariates. PROC LOGISTIC statistics related to the added exposure variable are summarized, including the creation of class variables for multi-level exposures if needed. The CMH option in PROC FREQ computes risk estimates for only 2x2 tables, and not for the more general rx2 tables. Our solution for rx2 level exposure variables is to create datasets for each comparison of an exposure l evel against the referent exposure group. PROC FREQ is run on these individual datasets and then re-run for the entire rx2 table to obtain the estimated overall linear association. The PROC LOGISTIC and PROC FREQ statistics are merged and stored in a datas et, using ODS output, to ease reporting the final summary set of statistics to the researcher. The macro is written to allow use within any operating system. Although the macro code is advanced, using the macro is simple for those with additional statistic al reporting needs.

Biography:
Stuart Long has a BS in Mathematics and Computer Science. He has programmed in SAS for 14 years. He worked with clinical studies involving Diabetes and Obesity for 9 years at the East Carolina School of Medicine. For the last 6 years, he has worked on epid emiological studies for the National Institute of Environmental Health Sciences in Druham North Carolina.

Rebecca Darden has an MS in biostatistics and an MPH in nutrition. She is a registered dietitian, and has worked in the field of epidemiology research and SAS programming for 15 years. Currently she is employed by WESTAT, on contract to the National Instit ute of Environmental Health Sciences, in Durham NC. She has worked on a variety of studies including ovarian cancer, neonatal disorder, colorectal cancer, kidney diseases, and lupus.