Correct way to interpret odds ratios

#1
I'm looking at a table of logistic regression results. Specifically, the table shows "Logistic regression model of hypertension status in relation to lead biomarkers in the normative aging study, Stratified by calcium intake".

The outcome variable is the presence of hypertension (1-yes, 0-no).

The covariates are: age>=70 (1/0), family history of hypertension (1/0), ever smoker (1/0), BMI (kg/m^2), blood lead, tibia bone lead levels, and patella bone levels. The results are stratified by calcium intake (Low calcium intake, and High Calcium intake).

My question is really a more simple, basic one: Why is it not the same to say the following two things:

(a) "odds of hypertension are 3 times higher in males aged>=70 WITH a family history of hypertension versus males <70 years without a family history of hypertension, adjusted for the other variables"

(b) "odds of hypertension are 3 times higher in males aged<70 WITHOUT a family history of hypertension versus males aged>=70 WITH a family history of hypertension, adjusted for the other variables"?

Please let me know if I can be clearer, and thanks so much in advance! A snapshot of the table is below: Screen Shot 2018-02-14 at 10.11.18 AM.png
 
Last edited:

hlsmith

Not a robit
#2
Are you just asking why you can't flip the ordering of the cases (2/3 not equal to 3/2, but are invertible 1/(2/3) = 3.2)?

Also, watch out you are changing two variable in the statements: age and family Hx, which I don't think you can derive from the table.