Skip to main content
. Author manuscript; available in PMC: 2017 Oct 1.
Published in final edited form as: J Thromb Haemost. 2016 Aug 17;14(10):1978–1987. doi: 10.1111/jth.13411

Figure 1. Graphical Illustration of time-varying exposure in the study.

Figure 1

Panel A: Time-varying model where outcome was assessed at the official date of disability pension. Subject A starts without venous thromboembolism (VTE), develops VTE during follow-up and receives subsequent disability pension. Subject B has no VTE during the entire study period, and contributes 15 years of non-exposed person time. Subject C has no VTE but receives disability pension. Subject D develops VTE during follow-up and dies before study end. Subject E develops VTE and receives disability pension within one year after the VTE.

Panel B: Time-varying model where outcome was assessed one year before the official date of disability pension. Subjects had to be on sick leave for at least one year before they could apply for disability pension. Therefore, subjects were likely to be disabled already one year before the disability pension was granted, and those who received disability pension within one year after the VTE were likely to be granted disability pension for reasons other than the VTE. By modelling the outcome date one year before the official disability date, the time-to-event was shortened by one year and subjects who were granted disability within one year after the VTE would be classified as non-exposed in this model (e.g. subject E).