| Popis: |
Purpose The Lung Allocation Score (LAS) has been used since 2005 to rank US lung transplant candidates for access to transplant. A revised LAS (LAS-R) system that will better predict outcomes for the current cohort of transplant patients has been approved and is awaiting implementation. Both the current LAS and LAS-R calculate a candidate’s wait-list mortality risk in the year post-listing and survival likelihood during first year posttransplant. We examined the impact of modifying LAS-R to use a long-term 5-year posttransplant survival model (LAS-R5). Methods and Materials SRTR standard analysis files were used to estimate posttransplant survival for lung recipients, 5/4/05-9/30/11. We calculated the LAS for all lung-only candidates aged ≥ 12 years on the waiting list 4/1/2011 using three models: current LAS, LAS-R, and LAS-R5. Using the same variables and parameter estimates for the LAS-R 5-year and 1-year models, we compared C statistics and Hosmer-Lemeshow tests to assess model fit. Results Boxplots show increased LAS-R for diagnosis group B compared with LAS, but LAS-R5 is similar to LAS-R. [ figure 1 ] C statistics for all models are comparable (LAS = 0.59, LAS-R = 0.61, LAS-R5 = 0.58), suggesting similar survival predictive power. However, the Hosmer-Lemeshow test (P = 0.0214) for the LAS-R5 indicates that observed mortality rates do not match expected event rates in subgroups of the model. Conclusions Using long-term 5-year survival with the LAS-R covariates does not improve LAS prediction of posttransplant survival likelihood. Possibly, because wait-list mortality risk is given twice the weight of posttransplant survival in LAS calculation, changes in posttransplant covariates do not significantly impact the LAS. |
