Hello,


I know there are many postings on this topic but I am still a bit confused.

I have cohort data with two consecutive pregnancies documented. I converted the data from long to wide. So for time-varying covariates, each id has a measurement related to 1st or 2nd pregnancy

I am trying to understand risk factor differences between women with recurrent Preeclampsia ( preeclampsia in pregnancy 1 AND preeclampsia in pregnancy 2) vs non-recurrent preeclampsia ( they only had the preeclampsia in pregnancy 2).

My hypothesis is that those with recurrent preeclampsia have a higher incidence (new diagnoses) of chronic hypertension, diabetes, and obesity prior to the second pregnancy compared to women with nonrecurrent preeclampsia.


my study population is all women with preeclampsia in first pregnancy
outcome/dependent variable is recurrent preeclampsia ; reference group is Non-recurrent preeclampsia
independent variables: new chronic hypertension, new obesity, new diabetes ( BEFORE second pregnancy)

I am not sure that I have the power to do anything besides a crude analysis, but if I did I ran into a question of whether to adjust for pregnancy one risk factors or pregnancy 2 risk factor ( example age)


Also and really the title of this post is, do I have to use GEE? Obviously the measurements are not independent. characteristics of 2nd pregnancy definitely are not independent of 1st pregnancy characteristics. I can adjust for pregnancy one characteristic, but Ithere is come clinical value in accounting for these factors. s it completely invalid to use logistic regression? If it is ok to, why is it ok?

I hope I am clear. Thank you!