3. (18 points) Effect measure modification occurs when the strength of an association varies according to the level of a third variable which is called an effect measure modifier.Confounding is the distortion of the association between an outcome of interest and an exposure because an important third variable has not been accounted for; this third variable is called a confounder.

The association between heavy alcohol consumption and the risk of oral cancer was investigated in a case-control study with 475 cases and 400 controls.The following results were seen:

Heavy Alcohol

Consumption Cases Controls

Yes 350 200

No 125 200

(a) (4 points) Calculate and interpret the odds ratio for this study.

(b) (8 points) Gender was considered a potential confounder and/or effect measure modifier in this study.The data were stratified by gender in order to assess these issues.Calculate and interpret the stratum-specific odds ratios among males and females using the following data:

Males

Heavy Alcohol

Consumption Cases Controls

Yes 300 150

No 50 50

Females

Heavy Alcohol

Consumption Cases Controls

Yes 50 50

No 75 150

(c) (3 points) Do you think gender is a confounder in this study?Explain your reasoning using the data and the definition of confounding given at the beginning of the problem.

(d) (3 points) Do you think gender is an effect measure modifier in this study?Explain your reasoning using the data and the definition of effect measure modification given at the beginning of the problem.