I'm conducting an observational study between two non-randomised treatment groups. I plan to balance the two using IPTW (inverse probability of treatment weighting). I'll be using some covariates (age, ethnicity, cancer stage, rurality etc.) to create the Propensity score first, before Weighting. My question is, do I have to include these covariates again in my final Regression model to assess overall or cancer-specific Survival between the two groups? [Haven't we already factored the covariates into the weights]?
I plan to use R for Overall Survival analysis (Kaplan-Meier and Multivariate Cox-regression) along with Cancer-specific survival analysis (Cumulative incidence Fine and Gray, and Cause-specific proportional hazards) for prognostic and etiological questions, respectively. I'll run the analysis on both un-weighted and weighted datasets.
P.S: Absolute novice here :)!
TIA
~Bewildered Doctor
