Dear all

Sorry if this topic has been previously addressed, but could not find it here or elsewhere.

Briefly, I'm studying risk of mortality in patients with liver cirrhosis. We want to evaluate the predictive capacity of two previously validated prediction models. One model estimates risk for 3-month mortality and one score estimates risk for 1- and 2-year mortality. There is censored data since some patients undergo liver transplantation, that significantly improves prognosis and changes the natural history. Normally, I would use a ROC curve to estimate the performance of these scores at specific time points, but here there is censored data. I then usually use C-statistic to evaluate a prediction model with time-to-event data, but in this setting I want to get data on how these scores work at specific points in time.

I.e., getting a C-index for score #1 to predict 3-month mortality in the presence of censored data, and a C-index for score #2 to predict 1- and 2-year mortality.

Any input on coding to use in this situation would be most appreciated.

best

Johannes