Left-truncated data with time-varying fitter

[benslack19]

Is it safe to assume that CoxTimeVaryingFitter supports left truncated data by simply having the first entry for a subject start at one's entry (e.g. W in the CoxPHFitter model)?

An example table would be like this:

id	start	stop	E
0	22	25	0
0	25	27	0
0	27	30	1
1	13	15	0
1	15	25	0

[benslack19]

Any thoughts on this @CamDavidsonPilon ?

[CamDavidsonPilon]

Sorry about the delay. Yea, I believe this is correct. However, do some tests!

[danielwiegand]

Hi @benslack19, @CamDavidsonPilon, does this mean that left truncation in time-varying survival regression is expressed simply by a subject's lowest start value being larger than 0?