Dear all,
I’m still learning. I feel I have learnt the essentials for cubic splines, cox regression and non parametric survival analysis. Now I’m at a stage where I’m trying to link everything together.
I want to see if there is an optimum number of surgical cases a surgeon needs to do to have better outcomes in gyanecology surgery, OUTCOME is reoperation.
My plan were these steps: - I don’t know if there is someone on here who is able to perhaps give me some insight/advice
1. Generate the number of cases done by the surgeon up until the date of the surgery for a patient - this remains as a continous various and prevents categorisation
2. Generate knots and create splines for surgical experience (should I use scatterplots to see the graphs?)
3. As my outcome is reoperation - I wanted to test using Cox Regression first, then test PH
4. I know no 3 will be violated and therefore will need to proceed to the non-linear method
Are my steps correct?
In no 3. I was going to look for the optimum model i.e using interaction terms as well. Once I decide on this model, can I continue to use this final model for no 4 ?
Do you think doing step 3 is a waste of time, if I know PH will be violated?
I’m still learning. I feel I have learnt the essentials for cubic splines, cox regression and non parametric survival analysis. Now I’m at a stage where I’m trying to link everything together.
I want to see if there is an optimum number of surgical cases a surgeon needs to do to have better outcomes in gyanecology surgery, OUTCOME is reoperation.
My plan were these steps: - I don’t know if there is someone on here who is able to perhaps give me some insight/advice
1. Generate the number of cases done by the surgeon up until the date of the surgery for a patient - this remains as a continous various and prevents categorisation
2. Generate knots and create splines for surgical experience (should I use scatterplots to see the graphs?)
3. As my outcome is reoperation - I wanted to test using Cox Regression first, then test PH
4. I know no 3 will be violated and therefore will need to proceed to the non-linear method
Are my steps correct?
In no 3. I was going to look for the optimum model i.e using interaction terms as well. Once I decide on this model, can I continue to use this final model for no 4 ?
Do you think doing step 3 is a waste of time, if I know PH will be violated?
