I have a sample of 1,000 observations, 100 of these observations had a history of stroke.
I plotted a survival curve for stroke, my indicator variable is history of stroke and my time variable is age when first had stroke onset. So for those with stroke, their time var = age when first had stroke onset. For those without stroke, their time var is censored at their current age at interview. This is easily done via SAS through PROC LIFETEST.
In my sample, patients that indicate that they had stroke were asked if any of their family members had a history of stroke or not. So only 100 with stroke were asked about the family history question.
## I wanted to see if the group with family history has an earlier onset of stroke or not. ##
Would it be incorrect if I were to do the following: first restrict my data to these 100 patients, and re-plot the survival curves by family history status?