*Purpose: The Scientific Registry of Transplant Recipients semi-annually assigns transplant programs to one of five performance tiers for 1-year graft survival in its program-specific reports (PSRs). Concerns about the historical variability in tier assignment over time have been expressed. However, historical analyses have two limitations: they cannot separate variability in program effects over time from random variability in outcomes, and they cannot compare the observed tier assignments to the correct tier assignments. To overcome these limitations, a simulation study was performed to assess the reliability of tier assignments when program effects are known and held constant.

*Methods: Simulation parameters approximated the cohort of adult recipients of deceased-donor kidneys in the January 2018 PSRs. One hundred PSR cohorts were simulated for 11,600 simulated programs (50 batches of 232) to produce 20 temporally independent tiers for each program.

*Results: The intraclass correlation for the 20 independent tier assignments was _0.380.39_0.40, which suggests that 39% of the overall variability in tier assignment was explained by the program. Intraclass correlations were _0.190.20_0.21, _0.340.35_0.36, and _0.560.57_0.58 for programs with 0-3, 3-10, and >10 expected events, respectively.

*Conclusions: A true tier was derived from the program effect for each simulated program. Figure 1 shows the simulated tier assignments by the number of expected events and the true tier. The assigned tier matched the true tier 37.5% of the time. The assigned tier matched the true tier 32.5%, 37.8%, and 44.4% of the time for programs with 0-3, 3-10, and >10 expected events, respectively. The assigned tier was within 1 tier of the true tier 83.6% of the time, and was within 1 tier of the true tier 79.0%, 83.7%, and 90.0% of the time for programs with 0-3, 3-10, and >10 expected events, respectively.

« Back to 2019 American Transplant Congress