I have used the rms package to create a nomogram for 1-year cardiovescular outcome after receiving breast cancer treatment based on the multivariable Cox proportional hazards model. However, I must admit that I do not fully understand the method and have been constantly challenged by my collaborators who are cardiologists without statistics background. Could anyone please help and rebut the following questions?
(1) Framingham risk scores established long time ago are easier to interpret than adding up the points on a nomogram. Is there a better way to visualize the probabilities of 1-year cardiovescular outcome?
(2) Why nomogram could be a better prediction tool compared to the risk score method from the statistician’s point of view and when it is translated into the clinical practice settings?
(3) Was the calibration plot generated from the validation cohort or the original dataset? How was the 45 degree line that indicates the ideal agreement calculated?
As a side note - the total sample size is 1400, and I did the internal validation with 200 bootstraps, no external validation.
Thank you all in advance.