In principle your computations were correct, but rounding errors can quickly become very big. The easiest and safest solution is to create a new variablegen age10 = (age -70)/10, and addage10instead ofage. Now your constant will be the odds of smoking for a 70 year old person with all otherxs set at 0, and the odds ratio forage10will be the ratio change for a 10 year increase in age.