Dear All,
Hi, I¢m trying to analyse some survival data, for which we have two treatment groups (treatment and control), and following a certain amount of time patients may take a further concomitant treatment, which in essence gives us four strata;
treatment + concomitant
treatment + no concomitant
control + concomitant
control + no concomitant.
Please note that the concomitant treatment cannot occur prior to 40 days.
However, the problem which I currently have is that when the data is analysed in the four strata, those patients who have an event prior to being able to take concomitant treatment (40 days) are only contributing to the ¡no concomitant¢ groups and this is significantly skewing the results. In essence what I would like to do is present just two survival curves up to this fixed time point, and then analyse the four strata after this. I am looking to be able to calculate both estimates of median survival and p-values for the treatment versus control, adjusted for concomitant medication, treatment difference. Does anyone know if this is possible and of a way to present this?
Many thanks,
Rhiannon
|