RESUMO
BACKGROUND: In France, the need for continuous monitoring of transplant center performance has recently become apparent. Cumulative sum (CUSUM) monitoring of transplantation is already been used to monitor transplant outcomes in the United Kingdom and in the United States. Because CUSUM monitoring can be applied by different methods, the objective was to assess and compare the performance of different CUSUM methods for detecting higher than expected (ie, excessive) graft failure rates. METHODS: Data come from the French transplant registry. Lung and kidney transplants in 2011-2013 constituted the control cohort, and those in 2014-2016 the observed cohort. The performance of CUSUM monitoring, according to center type and predefined control limits, was measured by simulation. The outcome monitored was 3-month graft failure. RESULTS: In a low-volume center with a low failure rate, 3 different types of control limits produced successful detection rates of excessive graft failures of 15%, 62%, and 73% and false alarm rates of 5%, 40%, and 52%, with 3, 1, and 1 excess failures necessary before a signal occurred. In a high-volume center with a high failure rate, successful detection rates were 83%, 93%, and 100% and false alarm rates were 5%, 16%, and 69%, with 6, 13, and 17 excess failures necessary before a signal occurred. CONCLUSIONS: CUSUM performances vary greatly depending on the type of control limit used. A new control limit set to maximize specificity and sensitivity of detection is an appropriate alternative to those commonly used. Continued attention is necessary for centers with characteristics making it difficult to obtain adequate sensitivity or sufficiently prompt response.