Budget Impact Model for Hypomethylating Agent Use for the Treatment of Myelodysplastic Syndromes (MDS)

Introduction. Current treatments for MDS include supportive care and the use of newer agents. The costs of supportive care and treatment with hypomethylating agents differ significantly. This study represents a budget impact model comparing the relative costs of supportive care and individual hypomethylating agents in this setting varying the model by average duration of treatment, time to response, and response rate.

Method. The P4 Healthcare database of over 1,300,000 lives was accessed for the period of January 1, 2007, to December 31, 2007, and includes key information such as dates of service, ICD-9 codes, procedure codes, costs associated with hospitalizations, and the payment amounts per date(s) of service. The average cost figures for the indicated procedures were determined from a Medicare (MC) and non-Medicare (nMC) payer perspective. Clinical response measures, side effects and other clinical event rates were established from published studies for the respective treatments. A 2-year budget impact model was developed that allows for evaluation of the various phases of treatment, including time to initial response and continued treatment due to ongoing clinical benefit. Each scenario includes estimated costs for drug therapy and administration, supportive care including transfusions, and costs associated with side effects such as febrile neutropenia and infections. Three potential scenarios were evaluated. The first analysis (Time to Response Model) established cost impact associated with initial time to response for the individual agents. This scenario used an initial estimate of time to response of 6 and 9 cycles for DAC and VID, respectively. The next scenario demonstrated the cost impact of ongoing treatment (Ongoing Treatment Model) for three treatment scenarios: Supportive Care (SC) only, Azacitidine (VID) followed by supportive care, Decitibine (DAC) followed by supportive care using the mean duration of response. The third analysis established the complete sequence of treatment (Complete Treatment Model) in which mean time to response, mean response rate and mean duration of response were used to model the overall impact comparing DAC and VID to SC for the entire 2-year period. The assumed overall response (complete response, partial response and hematologic improvement) rates were 73%, 60%, and 5% for DAC, VID, and SC respectively. Sensitivity analyses were conducted to demonstrate the impact based on differences in time to response, duration of response, and response rates. All analyses are reported by the MC and nMC perspectives.

Results. 1,929,974 remittance observations on 393,840 patients contributed to the analytic models. The Time to Response Model suggested that Dacogen (Mean MC ≈ $53,277; Mean nMC ≈ $56,464) results in lower cost than Vidaza (Mean MC ≈ $144,726; Mean nMC ≈ $140,322) relative to the time to response. Results of the Ongoing Treatment Model suggested that both treatment with Dacogen (Mean MC ≈ $88,795; Mean nMC ≈ $94,106) and Vidaza (Mean MC ≈ $176,887; Mean nMC ≈ $171,504) result in lower costs than supportive care alone (Mean MC ≈ $167,563; Mean nMC ≈ $174,225). The Complete Treatment Model suggests lower cost for the Dacogen regimen (Mean MC ≈$267,509; Mean nMC ≈ $280,512) over the Vidaza regimen (Mean MC ≈ $368,153; Mean nMC ≈ $365,183) and the Supportive Care regimen (Mean MC≈ $335,127; Mean nMC≈$348,451).

Discussion. The cost of SC for MDS is significant and interestingly the cost of hypomethylating agent treatment in this setting is less than SC. The budget impact model suggests there is utility in considering utilization of DAC as first line therapy over VID to minimize cost through the optimal application of the differential response rates and time to response rates.