Cost-effectiveness of anticoagulants for suspected heparin-induced thrombocytopenia in the United States

Abstract

Despite the availability of multiple nonheparin anticoagulants for the treatment of heparin-induced thrombocytopenia (HIT), few data are available comparing the cost-effectiveness of these agents. This analysis is particularly important when considering differences in the risk of adverse effects, routes of administration, requirements for phlebotomy and laboratory monitoring, and overall drug costs. We conducted a cost-effectiveness analysis of argatroban, bivalirudin, and fondaparinux for the treatment of suspected HIT from the institutional perspective. A 3-arm decision-tree model was developed that employs standard practices for anticoagulation monitoring. We incorporated published data on drug efficacy and probability of HIT-related thromboembolism and major bleeding. We considered both institutional costs and average wholesale price (AWP) and performed probabilistic sensitivity analyses (PSA) to address any uncertainty in model parameters. Using institutional costs, fondaparinux prevailed over both argatroban and bivalirudin in terms of cost ($151 vs $1250 and $1466, respectively) and adverse events averted (0.9989 vs 0.9957 and 0.9947, respectively). Results were consistent when AWP was used, with fondaparinux being less expensive ($555 vs $3081 and $2187, respectively) and more effective in terms of adverse events averted (0.9989 vs 0.9957 and 0.9947, respectively). The PSA confirmed our findings using both institutional costs and AWP. In conclusion, fondaparinux subcutaneous injection afforded significant advantages in terms of cost savings and adverse events averted compared with IV argatroban or bivalirudin infusions. Our data strongly suggest potential cost savings with fondaparinux and underscore the critical need for larger clinical studies of fondaparinux in the treatment of suspected HIT.

Figure 1.

Decision analysis tree The decision-analysis tree model starts when HIT is suspected and all forms of heparin are discontinued. Hypothetical patients are initiated on argatroban, bivalirudin or fondaparinux. The model assumes lab tests would be readily available after two days to confirm HIT diagnosis and continue treating patients with nonheparin anticoagulants for a total of 6 days or exclude the diagnosis of HIT and stop nonheparin anticoagulant after 2 days. The model assumed three expected outcomes while on nonheparin anticoagulant: HIT related VTE, major bleeding, or no complication. The baseline rates of outcomes reported in studies and the calculated probabilities of these outcomes can be seen in this figure.

Figure 2.

Probabilistic sensitivity analysis using institutional drug costs Two thousand simulation samples were used to confirm analysis using institutional costs. Y-axis represents the incremental cost and X-axis represents incremental adverse event averted. Samples in north quadrants indicate more expensive drug while samples in south quadrants indicate more cost-savings. East quadrants represent higher rate of adverse events averted, and samples in west quadrants mean lower rate of adverse events averted.

Figure 3.

Probabilistic sensitivity analysis using average wholesale prices Two thousand simulation samples were used to confirm analysis using average wholesale prices. Y-axis represents the incremental cost and X-axis represents incremental adverse event averted. Samples in north quadrants indicate more expensive drug while samples in south quadrants indicate more cost-savings. East quadrants represent higher rate of adverse events averted, and samples in west quadrants mean lower rate of adverse events averted.