Program-Effect Attenuation With Time Posttransplant

N. Salkowski,¹ L. Hunsicker,^1,2 J. Snyder,¹ B. Kasiske.^1,3

¹Scientific Registry of Transplant Recipients, Minneapolis Medical Research Foundation, Minneapolis, MN
²Department of Medicine, University of Iowa, College of Medicine, Iowa City, IA
³Department of Medicine, University of Minnesota, Minneapolis, MN.

Meeting: 2015 American Transplant Congress

Abstract number: D241

Keywords: Kidney transplantation, Methodology, Public policy

The program-specific reports produced by SRTR include 1-month, 1-year, and 3-year posttransplant graft and patient survival measures. The 1-month and 1-year measures use the same 2.5-year cohort of transplant recipients. The 3-year measures use an older non-overlapping 2.5-year cohort. The 3-year measures have limitations: they are calculated for older cohorts, so their relevance for current program evaluation is suspect. Follow-up beyond 3 years is ignored. Graft failure and patient death events that occurred during the first year posttransplant are included in the 3-year measures, so 3-year measures are not distinct from 1-year measures. Separate 5-year conditional survival models were built for deceased and living donor adult kidney graft and patient survival using the 3-year model cohorts, excluding recipients with first-year events to avoid the limitations of the 3-year models. For each model, a hazard ratio (HR) was calculated for each program as a performance measure. HRs tended to be least well correlated between the 1-year and 5-year models (Table 1). There was weak correlation between 1-year and 5-year estimated program HRs for deceased donor adult graft survival (Fig. 1), and substantial program-level variability in 5-year conditional graft survival. Although how much long-term outcomes are determined by factors controlled by transplant programs is unclear, these results suggest that 5-year conditional survival metrics could be a useful supplement to 1-year survival metrics.

Correlations between estimated program hazard ratios

	Deceased Donor		Living Donor
Adult Kidney Models	Graft	Patient	Graft	Patient
1-Year, 3-Year	0.24	0.20	-0.03	0.00
3-Year, 5-Year	0.63	0.66	0.61	0.70
1-Year, 5-Year	0.13	0.06	-0.01	0.00

Figure 1. Scatterplot of estimated program hazard ratios for 1-year and conditional 5-year deceased donor adult graft survival. "X" symbols, programs with unusually low 5-year conditional graft survival; "+" symbols, programs with unusually low 1-year graft survival.

« Back to 2015 American Transplant Congress