Odds ratio displayed 'per 10 units of predictor'

Dear all,

In SPSS, I performed a linear regression with 'change in blood concentration of mineral X' as (continuous) outcome, and age as (continuous) predictor. Both the outcome and the predictor are normally distributed. The Beta (95% CI) is 0.007 (0.001, 0.013). To express the beta for 'change in blood concentration of mineral X' per 10 years increase, I multiplied both the Beta and the 95% CI by 10. I think this is acceptable.

Now I want to do more or less the same for the OR (95% CI) in a logistic regression with a dichotomous outcome (disease or no disease) with again age as risk factor (continuous variable). The logistic regression in SPSS yields an OR (95% CI) of 1.081 (1.022, 1.144). Is the correct way to display this as 'per 10 years increase' by doing 1.081^10 (1.022^10, 1.144^10)? I.e. convert everything to the power of 10?

I couldn't find the answer to this question online, but by the idea I have of OR's, I think this is the correct way. Can anyone either approve this, disprove this, or show me an explication where I can find this out by myself?

Thanks in advance!


Less is more. Stay pure. Stay poor.
As for the logistic, that sounds right and it can be easily checked my dividing the independent variable by 10 and rerunning the model.

However, the linear example you provided feels off. CI's are beta(*1) +/- (1.96*SE), so wouldn't 10 unit change be beta_estimate(*10) +/- (1.96*SE)?