Regression - correcting for the same covariate with a different value twice

Question

[ika]

I want to perform a regression to check the correlation between two phenotypes. Both phenotypes are age-dependent, so I need to correct for this.

There is one problem, however: the two phenotypes were not recorded at the same time and I don't have data available for the same age per individual. Also, the time span between the two records differs between individuals.

Basically, I have the following data:

• phenotype A
• age when phenotype A was recorded
• phenotype B
• age when phenotype B was recorded

And I want to perform regression of phenotype A on phenotype B, while correcting for age. I'm a bit unsure how to go about this.

My thought was, maybe I should regress phenotype B on 'age when phenotype B was recorded' first, then use those residuals in the regression while correcting for 'age when phenotype A was recorded'. Would this be a correct way to go about this?

I appreciate any input!

[rpolicastro]

If there are more than 5 factor levels for age, you could perhaps try adding age as a random effect.