Offer Acceptance for Liver, Heart, and Lung Programs, and New Posttransplant Outcomes Models for Liver Programs
To identify the organ utilization practices of transplant programs, SRTR is working toward including program-specific offer acceptance practices in the program-specific reports (PSRs). Kidney offer acceptance was integrated into the PSRs in July 2017. Since then, SRTR has developed liver, heart and lung offer acceptance models and plans to integrate offer acceptance reports for these organs into the January 2018 PSRs. The models are being previewed on our website.
SRTR has also developed new liver posttransplant models for graft and patient survival, and plans to integrate these models into the January 2018 PSRs. These models are the first liver models to use the penalized modeling framework (LASSO) that has previously been applied to kidney, heart, and lung posttransplant models. This modeling framework allows SRTR to consider a large number of potential predictors without overfitting.
The new liver models also change how missing data is handled. Like the kidney, heart, and lung posttransplant models, when the models are fit, missing data is handled through multiple imputation. In multiple imputation, the non-missing data is used to predict the values of the missing data multiple times, producing multiple similar – but not identical – data sets with all of the missing data replaced with predicted values. The fitted coefficient values are then averaged across the different models.
Although multiple imputation works well for fitting the models, it can't be used for program evaluations. To encourage programs to submit data completely, SRTR calculates the lowest risk associated with the non-missing values of each predictor, then assigns that lowest-risk value to the missing data effect. Since missing data is treated as equivalent to the lowest-risk data, there is no incentive for programs to leave data elements missing.
The new liver models can be previewed here. More detailed information about the modeling process can be found in the following publication: Snyder JJ, Salkowski N, Kim SJ, Zaun D, Xiong H, Israni AK, Kasiske BL. Developing statistical models to assess transplant outcomes using national registries: The process in the United States Transplantation. 2016;100:288-294.
Comments on the offer acceptance and liver models can be submitted to firstname.lastname@example.org.