Simulated data used in "The importance of censoring in competing risks analysis of the subdistribution hazard"

The simulated data used for analysis in "The importance of censoring in competing risks analysis of the subdistribution hazard". Simulated using the method described in Additional file 1.

Warning: Large file. Contains 1000 datasets of 300 observations each, for each of 105 parameter combinations (31,500,000 rows). Some programs (e.g. Excel) will not be able to open it in full.

csv file with columns:

p.comp: risk of the competing event in exposure group A for this scenario [0 to 0.30 in increments of 0.05]
lnb.cens: log(hazard ratio) for loss to follow-up in old versus young individuals for this scenario [0 to 1 in increments of 0.25]
lnb.evt: log(subdistribution hazard ratio) for the event of interest in exposure group B vs group A for this scenario [0, 0.5, 1]
sim: ID of the simulated dataset for this scenario [1-1000]
exposure: exposure group (0 = A, 1 = B) of this individual
age: age group (0 = young, 1 = old) of this individual
time: time-to-event or censoring for this individual
evtcode: event type (0 = censoring, 1 = event of interest, 2 = competing event)
censcode: type of censoring (1 = end-of-study, 2 = loss to follow-up)