ST02
A SAS® Program for the Computation of the Crude, Stratified and Mantel-Haenszel Odds Ratio in Case-Control Study Data Analysis of Nx2xK tables.
Contributed
Ilene Brill
Department of Epidemiology and International Health, University of Alabama at Birmingham
,Dr. Fabio Barbone
Department of Epidemiology and International Health, University of Alabama at Birmingham Abstract:
For the analysis of epidemiologic case-control studies, a SAS macro program was constructed to compute estimates of the exposure odds ratio (OR) (ratio of exposed to unexposed cases divided by the ratio of exposed to unexposed noncases) beyond the standard
2x2 table. For each exposure level compared to the referent zero exposure level, the program calculates crude ORs, stratified ORs based on all possible combinations of confounders and the Mantel-Haenszel estimator of the OR adjusted for confounding. In ep
idemiology confounding, defined as the mixing of the effect of another factor (the confounder) associated with both the outcome and the exposure, is a major threat to validity. The program accommodates varying number of confounders and levels for each conf
ounder, one response variable with two levels (case/control) and one exposure with varying number of levels. The user designates as arguments to the macro: temporary input data set name, name prefix for the confounders PREFIX1-PREFIXn, outcome and exposure
variable names, number of exposure levels and the variable name for the weights assigned to the frequency values. The epidemiology instructor preferred using these methods of computing ORs over logistic regression to lead students to a better understandin
g of the relation among the contributing variables.
Biography:
Ilene Brill has been a Systems Analyst in the Department of Epidemiology and International Health, School of Public Health, University of Alabama at Birmingham (UAB) for 11 years using SAS for Windows for data management and analysis of STD epidemiology pr
ospective follow-up studies and occupational epidemiology retrospective follow-up and case-control studies. Prior work experience as a SAS programmer include: Department of Epidemiology and Biostatistics, University of Tromsų, Tromsų, Norway; Channing Labo
ratory, Boston, MA; Department of Epidemiology, Massachusetts Eye and Ear Infirmary; Department of Health Care Organization and Policy, UAB.M.P.H. Biometry, University of Alabama at Birmingham; B.A. Biology, University of Pennsylvania.
Dr. Barbone is an Associate Professor in the Department of Epidemiology and International Health, School of Public Health, University of Alabama at Birmingham. Dr. Barbone, a physician and doctoral-level epidemiologist, applies epidemiologic methodology to
the study of the natural history of disease and injury. With International experience in cancer, he hasconducted fundamental work on air pollution and lung cancer. He also hasdirected research for the purpose of disentangling the relative role ofsocioecon
omic factors, migration, occupational exposure, and life-stylefactors in the etiology and survival of cancer of the lung, breast, colon,oral cavity, etc. Dr. Barbone is also involved in International studies ofinjury, particularly case-crossover studies of
childhood injury and roadtraffic accidents. Currently, he is developing a methodologic frame andacademic program for the application of epidemiology to the study of adverse effects of drugs and of errors in medicine.