Fig. 1

Illustration of the “from registration” and “from cross-section” approaches to modelling waitlist mortality. For revision of MELD, typically 90-day time-stopped Cox models are used. The “from registration” approach (left) uses time since registration as the time scale and adjusts for biomarkers reported at registration. The “from cross-section” approach (right) models time-until-death from cross-section dates, pre-specified in calendar time. Adjustment is for MELD biomarkers reported before the cross-section date