Indirect and direct standardization for evaluating transplant centers

Abstract

To assess the quality of health care, patient outcomes associated with medical providers are routinely monitored in order to identify poor (or excellent) provider performance. To avoid confounding by risk factors, both indirect and direct standardization have been used for comparing outcome rates or prevalence for different providers. There has been an ongoing debate as to which standardization method is more appropriate. To compare the performance of indirect and direct standardization for the purpose of ranking transplant centers, we analyzed post-transplant mortality using the national kidney transplant data. Included in our analysis were 116,601 patients (from 230 transplant centers) who underwent kidney transplantation between January 2006 and December 2012. Multivariate logistic regression model was used to model the 30-day mortality, which were estimates of failures (grant failure or death) in the 30 days after the transplant surgery. Concordance indexes, kappa coefficients and Spearman’s rank correlation coefficient were computed. The estimated values from these statistics for the indirect standardized method were similar to the direct standardization. The results suggest that both indirect and direct standardized methods provide similar ability to distinguish center effects.