Cost-Effectiveness Model of Cytomegalovirus Management Strategies in Renal Transplantation: Comparing Valaciclovir Prophylaxis with Current Practice

Background: Cytomegalovirus (CMV) disease may occur following renal transplantation and has been shown to have health and cost consequences in this setting. Objective: To compare the cost effectiveness of different CMV management strategies for renal transplant patients: prophylaxis with (i) oral valaciclovir or (ii) intravenous ganciclovir; viral testing for CMV followed by (iii) pre-emptive therapy with intravenous ganciclovir or (iv) adjustment of immunosuppression and intensive monitoring; or (v) waiting to treat when CMV disease develops. Methods: A decision-tree model was constructed that included the different management strategies for the donor seropositive/recipient seronegative (D+R-) population. Clinical outcomes for the D+R- population came from clinical trials. Treatment algorithms and costs for CMV syndrome and tissue invasive disease were developed from published literature and UK physician interviews. One- and 2-way sensitivity analyses were performed. Study Perspective: UK National Health Service. Results: Prophylaxis with either oral valaciclovir or intravenous ganciclovir dominated (lower costs and fewer cases of CMV disease) the pre-emptive treatment and wait-and-treat strategies. The cost per patient was from Lstg 157 to Lstg 438 higher with oral valaciclovir prophylaxis compared with intravenous ganciclovir prophylaxis and the incremental cost per case of CMV disease avoided with valaciclovir prophylaxis ranged from Lstg 2243 to Lstg 8111 (1996 values). These results are sensitive to the efficacy of intravenous ganciclovir prophylaxis and CMV management costs. Conclusions: For D+R- renal transplant patients, prophylaxis is the dominant (more effective and less costly) management strategy compared with pre-emptive and wait-and-treat strategies. The cost per patient with oral valaciclovir prophylaxis compared with intravenous ganciclovir prophylaxis is slightly higher in our base case scenario, but may be lower under reasonable alternative assumptions.