Online Program

Return to main conference page

Wednesday, January 10
Wed, Jan 10, 5:30 PM - 7:00 PM
Crystal Ballroom CD & Prefunction
Welcome Reception & Poster Session I

Statistical Hypothesis Testing Fails to Categorize Transplant Program Performance (304178)

Nicholas Salkowski, Scientific Registry of Transplant Recipients 
Jon Snyder, Scientific Registry of Transplant Recipients 
*Andrew Wey, Scientific Registry of Transplant Recipients 

Keywords: Public reporting, rating system

In the United States, the Scientific Registry of Transplant Recipients publishes public reports on transplant program outcomes. To inform healthcare consumers, programs have been categorized as "as expected," "better than expected," or "worse than expected" based on the outcome of a statistical hypothesis test. The hazard ratio for 1-year posttransplant graft survival is used for posttransplant outcomes. However, the statistical hypothesis test is underpowered for most transplant programs due to relatively small effective sample sizes (e.g., < 10 expected graft failures for most programs). The low power causes poor differentiation of graft survival in the "as expected" group, e.g., a 4-fold difference in hazard ratios. In contrast, the transplant rate ratio is used for pretransplant outcomes and has a 10-fold increase in effective sample size. The higher power causes poor differentiation in the "better than expected" and "worse than expected" groups, e.g., a 5-fold difference in transplant rate ratios. Thus, statistical hypothesis tests fail to group programs with similar performance levels and may fail to inform healthcare consumers of program performance.