Session: 902. Health Services Research—Malignant Conditions (Lymphoid Disease): Poster I
Hematology Disease Topics & Pathways:
Methods: We used a Markov model to estimate strategy costs and quality-adjusted life years (QALYs) for two patient scenarios: a 3-year-old with standard-risk ALL receiving 2 asparaginase doses, and a 15-year-old with high-risk ALL receiving 7 asparaginase doses over a 5-year time horizon. Patients entering the model received premedication with serum asparaginase level monitoring, monitoring only, or no premedication/monitoring. Literature data were used for hypersensitivity reaction and silent inactivation risks following each asparaginase dose. Silent inactivation was not identified the non-monitoring strategy. Disease outcomes, therapy and associated additional care costs, and health state quality-of-life utilities were obtained from the literature and US databases. Evaluation took the societal perspective, with costs and effectiveness discounted at 3%/yr. Multiple sensitivity analyses were performed.
Results: In both the standard-risk and high-risk analyses, premedication was the least costly strategy. In the standard-risk model, premedication with monitoring cost $4,586 less than monitoring alone, resulted in 8% fewer changes to Erwinia and 0.01 additional QALYs. It cost $1,993 less than no premedication/monitoring, resulted in 3% fewer changes and 0.08 additional QALYs. In the high-risk model, premedication cost $29,757 less than monitoring alone, resulted in 7% fewer medication changes and 0.01 fewer QALYS; thus, monitoring alone was expensive, costing >$2 million/QALY gained compared to premedication and monitoring. Premedication cost $11,255 less than no premedication/monitoring, resulted in 2% fewer changes and 0.07 additional QALYs. Individual variation of all model inputs did not change the favorability of premedication and monitoring for either model. In probabilistic sensitivity analyses varying all parameters simultaneously over distributions 1000 times, premedication and monitoring was favored in >86% of model iterations in both standard- and high-risk scenarios.
Disclosures: No relevant conflicts of interest to declare.
See more of: Oral and Poster Abstracts