To answer your question: Basically, Yes
With two year intervals, the method recasts the data into a series of two-year followup studies; the baseline covariates are those known at the start of each study. The Poisson model for for interval \(i\) is equivalent to a Cox model with (unknown) constant hazard \(\lambda_i\) in the interval. The coefficients for the covariates can change from interval to interval. Therefore, the model does not assume proportional hazards. As a subject can start followup in multiple intervals, standard errors cluster on the subject..
With two year intervals, the method recasts the data into a series of two-year followup studies; the baseline covariates are those known at the start of each study. The Poisson model for for interval \(i\) is equivalent to a Cox model with (unknown) constant hazard \(\lambda_i\) in the interval. The coefficients for the covariates can change from interval to interval. Therefore, the model does not assume proportional hazards. As a subject can start followup in multiple intervals, standard errors cluster on the subject..
Comment