Evaluation of Flagging Criteria of United States Kidney Transplant Center Performance: How to Best Define Outliers?. Issue 6 (June 2017)

• Record Type: Journal Article
• Authors: Schold, Jesse D.; Miller, Charles M.; Henry, Mitchell L.; Buccini, Laura D.; Flechner, Stuart M.; Goldfarb, David A.; Poggio, Emilio D.; Andreoni, Kenneth A.
• Is Part Of: Transplantation. Volume 101:Issue 6(2017)
• Publication Date: 2017-06
• Subjects: Transplantation of organs, tissues, etc -- Periodicals; Transplantation immunology -- Periodicals; 617.95
• DOI: 10.1097/TP.0000000000001373
• Languages: English
• ISSNs: 0041-1337

Abstract

Abstract : Background: Scientific Registry of Transplant Recipients report cards of US organ transplant center performance are publicly available and used for quality oversight. Low center performance (LP) evaluations are associated with changes in practice including reduced transplant rates and increased waitlist removals. In 2014, Scientific Registry of Transplant Recipients implemented new Bayesian methodology to evaluate performance which was not adopted by Center for Medicare and Medicaid Services (CMS). In May 2016, CMS altered their performance criteria, reducing the likelihood of LP evaluations. Methods: Our aims were to evaluate incidence, survival rates, and volume of LP centers with Bayesian, historical (old-CMS) and new-CMS criteria using 6 consecutive program-specific reports (PSR), January 2013 to July 2015 among adult kidney transplant centers. Results: Bayesian, old-CMS and new-CMS criteria identified 13.4%, 8.3%, and 6.1% LP PSRs, respectively. Over the 3-year period, 31.9% (Bayesian), 23.4% (old-CMS), and 19.8% (new-CMS) of centers had 1 or more LP evaluation. For small centers (<83 transplants/PSR), there were 4-fold additional LP evaluations (52 vs 13 PSRs) for 1-year mortality with Bayesian versus new-CMS criteria. For large centers (>183 transplants/PSR), there were 3-fold additional LP evaluations for 1-year mortality with Bayesian versus new-CMS criteria with median differences in observed and expected patient survival of −1.6% and −2.2%, respectively. Conclusions: A significant proportion of kidney transplant centers are identified as low performing with relatively small survival differences compared with expected. Bayesian criteria have significantly higher flagging rates and new-CMS criteria modestly reduce flagging. Critical appraisal of performance criteria is needed to assess whether quality oversight is meeting intended goals and whether further modifications could reduce risk aversion, more efficiently allocate resources, and increase transplant opportunities. Abstract : The authors evaluate incidence, survival rates and volume of low performance centers with Bayesian, Old- and New-CMS criteria among US kidney transplant programs. The incidence of flagging and difference in observed and expected outcomes are significantly different by performance criteria.