Hello there, I have the data: number of cases of disease A and disease B by age in a population and the census of that population. I also know which cases present with both diseases. My hypothesis is that both diseases are not related so I'm doing an equivalence test for the risk of having disease B if you have disease A. Mainly: B+A+/A+ = B+A-/A- (the margin is the 20% of the incidence in the A- group). What I have so far is: proc freq data = datas order=data;
tables A*B/norow nocol nopercent riskdiff(equiv margin=0.0002);
weight counts;
run; I'm using counts because I have number of cases in the population and total population size, so I'm estimating the non diseased population from those numbers. My problem is that the probability for both diseases increases with age, so I need to to remove the age factor from my analysis. Is there a way to remove the co-variate from this model? Because I was thinking about standardization, but all the data comes from the same population. Any thoughts?
... View more