Background: After kidney transplantation, early readmission is independently associated with graft loss and mortality. The mechanism of this association is poorly understood. Understanding the timeline of risk, that is, during the readmission hospitalization versus periods postreadmission, will provide additional insights.
Methods: We used national registry data to study 56 076 adult Medicare-primary first-time kidney transplant recipients from December 1999 to October 2011. Piecewise Cox proportional hazard models were used to estimate the association between graft loss, mortality, and readmission for 2 periods: readmission hospitalization and postreadmission.
Results: During the readmission hospitalization, graft loss was substantially higher (deceased donor kidney transplant [DDKT] without delayed graft function [DGF] hazard ratio: 24.634.447.9, P < 0.001; with DGF: 10.815.221.4, P < 0.001; live donor kidney transplant [LDKT]: 18.136.774.2, P < 0.001) and mortality was substantially higher (DDKT without DGF: 14.120.830.7, P < 0.001; with DGF: 9.0312.818.0, P < 0.001; LDKT: 9.0018.241.3, P < 0.001). Immediately after readmission discharge, graft loss (DDKT without DGF: 2.082.402.77, P < 0.001; with DGF: 1.832.142.51, P < 0.001; LDKT: 2.002.503.13, P < 0.001), and mortality (DDKT without DGF: 2.162.432.73, P < 0.001; with DGF: 1.832.162.88, P < 0.001; LDKT: 1.902.342.88, P < 0.001) remained elevated, but much less so. After readmission, the hazard of graft loss remained, but decreased 19% per year for DDKT recipients (time varying coefficient 0.780.810.85, P < 0.001) and 14% per year for LDKT recipients (0.790.860.93, P < 0.001). The hazard of mortality remained, but decreased 14% per year for DDKT recipients (0.830.860.89, P < 0.001) and 9% per year for LDKT recipients (0.850.910.98, P < 0.001).
Conclusions: In conclusion, readmission is most strongly associated with graft loss and mortality during the readmission hospitalization, but also portends a lasting, albeit attenuated, risk postreadmission.