Question: Use Proc GLM to fit an appropriate model that compares mean BMI for people

Question 1 [10 marks]

(a)  Use Proc GLM to fit an appropriate model that compares mean BMI for people with and without DIABETES, providing an estimate of the difference in mean BMI and 95% confidence interval for the difference between people with and without diabetes.

(b) Use Proc GLM to fit an appropriate model that compares mean BMI for people with and without DIABETES after adjustment for SEX, AGE and EXERCISE (include curvature terms for AGE and EXERCISE if needed - justifying why if you do). Obtain and interpret the p-value, estimate and 95% confidence interval for the difference in (adjusted) mean BMI for people with and without diabetes.

 (c) Write down an algebraic expression of the fitted model from and provide an interpretation of the effects of AGE, SEX and EXERCISE on mean BMI.

(d) Succinctly (in a sentence or two) describe the differences in the comparison of mean BMI between diabetes and non-diabetes from the models fit in (a) and (b), providing a summary of the impact of the confounders fitted in.

 (e) Use Proc GLM to fit an appropriate model that tests if the difference in mean BMI for people with and without DIABETES is the same for males and females (after adjustment for, AGE and EXERCISE as determined from (b). Obtain the estimates of all beta coefficients and provide an interpretation of each beta coefficient associated with either SEX or DIABETES.

 (f) Using your model from (e), provide a one sentence conclusion about the impact of SEX on the relationship between mean BMI and DIABETES.

 Question 2 [10 marks]

(a) Use Proc GLM to fit linear model that compares mean DBP for adults between those who are and are not on treatment for hypertension (use RXHYPER as a categorical variable). Obtain the estimated differences in mean DBP for adults between the groups, calculating a 95% confidence interval for this difference.

 (b) Investigate the impact of BMI (as a continuous variable with linear term only) on the relationship between RXHYPER (as a categorical variable) and DBP using two separate approaches: Consider BMI as a confounder of the relationship between RXHYPER and DBP. Consider BMI as an effect modifier of the relationship between RXHYP

DRAFT/STUDY TIPS