linear mixed model with different entry date

[pierre martin]

Hi everyone,
I am interested in assessing the association of blood pressure levels of control after a diagnosis of hypertension with cognitive decline. I was thinking about using linear mixed models but the problem is that my exposure (level of control of BP) is not at baseline for everyone! Participants will be included once they have a first diagnosis of hypertension and then we will assess the association of their BP level at the time of first diagnosis with subsquent cognitive decline.

I think linear mixed models are appropriate to answer this question in this case but I just wanted to make sure becore writing my STATA code.

Any help/advice will be highly appreciated! Thank you so much!
Best,
Pierre

[Carlo Lazzaro]

Pierre:
there's another issue here.
Linear -mixed- models are close cousin of -xtreg, mle-.
One requirement of panel data is the repeated measurements of the same variables on the same sample of units (let's forget attrition for a while) at (more or less) equally spaced intervals.
From your description, a pooled OLS looks more reasonable.

[pierre martin]

Hi Carlo,

Thank you so much for your reply. I will read more about it then.

Best regards,
Pierre