Adjust For Age In Case/Control Study?

Question

[Dima Lvov]

We have a set of cases/controls for stroke, with fe loci genotyped and also age. We found few allelic combinations associated with the disease, but would like to account for age also. Some publications state that their logistic regression p-values are age-adjusted, and we now wonder what it means.

Do they simply include age in their regression model or is there a special procedure for that?

Searched all for a week and still nothing sensible, any advice appreciated.

Answer 1

[Darren J. Fitzpatrick]

Hi Dima,

You can include age as a covariate in your regression model such that the model will look like

Stoke ~ Age + SNP

It is as simple as that really!

Good Luck!

[Darren J. Fitzpatrick]

@Dima - It is possible that in your data set age is not an appropriate covariate. Try a simple regression analysis (Stroke ~ Age) and check if the beta coefficient is significant overall. If it is, include it as a covariate - if not, omit it.

[Dima Lvov]

Thank you Darren and Larry, we tried different tools for regression analysis, and what looks suspicious that this "adjustment" (adding age as covariate) affects the regression coefficient for the SNPs only very little (less then one percent). What is affected is the p-value of that coefficient. Would you say its normal?

[xliu4255]

Hi, when trying to use age of onset as a covariate for a disease susceptiblity case-control study, does the "age of onset" mean in the control group? Simply the time when the data is collected? Or does this kind of study is not suitable to use age of onset as a covariate?

Answer 2

[Larry_Parnell]

Yes, it is important to adjust for age when dealing with an age-related disease such as hypertension and stroke. Age is included in the regression model. Try this and if you still have questions, send me a direct message.