Thank you very much for the code and detailed explanation. The manuscript is superb. I am working on an external validation study and will be citing both the paper as well as your GitHub site.
Please provide the code for calibration metrics calculated using the sub distribution hazard approach by Austin et al.
Thank you.
The standard error used from the Debray paper seems to be the SE of log(O/E), in which case I believe the confidence intervals should be: "lower" = exp(log(OE) - qnorm(0.975) * aj$se / aj$obs) "upper" = exp(log(OE) + qnorm(0.975) * aj$se / aj$obs)
Thanks for considering my comment. I appreciate the thorough guide to this area!