A total of 126 women were randomly selected from the Connecticut cancer registry to participate in this study of the association between estrogen use and the incidence of endometrial cancer. Controls were matched (1:1) to cases within one year of the age at diagnosis, marital status, and living in the same retirement community at the time of diagnosis.
Covariates included self-reported lifetime history of hypertension, gallbladder disease, and non-estrogen drug use.
Data file: fexamq4 (N=126) The variables are as follows: ID - identification variables; matched pairs have the same ID Case - case status, 1=case, 0=control Age - age in years Est - estrogen use, 1=yes, 0=no Gall - history of gallbladder disease, 1=yes, 0=no Hyper - history of hypertension, 1=yes, 0=no Nonest - non-estrogen drug use, 1=yes, 0=no Use multiple regression analysis to examine the association between estrogen use and the incidence of endometrial cancer. Examine if the matching worked. Include all covariates in the standard model. Do not test for any interactions. Identify the "best" model based on validity and precision. Evaluate and discuss the impact of selection bias, misclassification, and confounding on the results. Briefly describe the data analysis plan and the results.