Since the calculation of binary exposures, Levine formula is introduced. But how to calculate when the study has multi categorical exposure and confounding variables
There is no easy way. You may have to look at each category separately to determine effect. Confounding makes it worse. You are entering the territory of a potential Simpson's paradox. See the attached.
The causal-pie modeling techniques (below) may prove useful:
Liao SF, Lee WC*. Sufficient cause modeling with matched data using SAS. Epidemiology 2013;24:936-937.
Liao SF, Lee WC*, Chen HC, Chuang LC, Pan MH, Chen CJ. Roles of human papillomavirus infection, high vaginal parity, and their interaction on cervical cancer risks after a follow-up of more than 10 years. Cancer Cause Cont 2012;23:703-708.
Liao SF, Yang HI, Lee MH, Chen CJ, Lee WC*. Fifteen-year population attributable fractions and causal pies of risk factors for newly developed hepatocellular carcinomas in 11,801 men in Taiwan.PLoS ONE 2012;7(4):e34770.doi:10.1371/journal.pone.0034779.
Lee WC*. Completion potentials of sufficient component causes. Epidemiology 2012;23:446-453.
Liao SF, Lee WC*. Weighing the causal pies in case-control studies. Ann Epidemiol 2010;20:568-573.
Logistic regression is a powerful tool to analyze data where you have multiple categorical variables. What you come out with is an odds ratio, controlling for the various variables you enter. This provides you with the opportunity to look at the effects of the variables of interest controlling for each of the others. Common statistical technique in epidemiological studies. This does not yield a direct measure of attributable risk, however. .But you can compare risks for groups with different characteristics. The references by Wen-Chung Lee and Joseph Alvarez are excellent ones and should be very helpful.
Its depend on which type of article: prevalence, case vs control, experimental etc.
- Badges
- Science topic