Cost–Effectiveness of Rivaroxaban Versus Heparins for Prevention of Venous Thromboembolism After Total Hip or Knee Surgery in Sweden

Lars Ryttberg; Alex Diamantopoulos; Fiona Forster; Michael Lees; Anina Fraschke; Ingela Björholt


Expert Rev Pharmacoeconomics Outcomes Res. 2011;11(5):601-615. 

In This Article

Expert Commentary

The present analysis assessed the cost–effectiveness of a new pharmacological agent, rivaroxaban, and extrapolates results in two phases: medium term (90 days) and long term (5 years), as recommended by Sullivan et al.[25] and similar to Gordois et al.,[26] Lundkvist et al.[6,7] and Sullivan et al.[27] This ensured that all relevant health-related impairment was incorporated in the analysis together with the acute-phase events.

In THR surgery patients, an extended duration of rivaroxaban prophylaxis versus LMWH resulted in an incremental cost–effectiveness ratio of SEK30,000–36,000 per QALY. In TKR surgery patients, rivaroxaban was associated with average cost savings of SEK873 per patient, as well as improved health outcomes and, therefore, dominates LMWHs.

Multiple economic evaluations,[26–30] including from Sweden[5,6] and Norway,[31] have evaluated the cost–effectiveness of VTE prophylaxis based on clinical trial results. The results of the present analysis are very similar to those from Lundkvist et al.,[6] who compared fondaparinux and enoxaparin. In the THR population, the total cost of enoxaparin was marginally different from that estimated by Lundkvist et al.;[6] SEK2313 versus approximately SEK2400 (cost is presented in SEK based on the exchange rate used by Lundkvist and colleagues:[6] 9.1 Swedish crowns to €1). In the TKR population, the cost estimates were lower compared with those of Lundkvist et al.[6] (SEK2329 vs SEK2800), but within a margin of error acceptability given the methodological differences of the two analyses. Moreover, the estimated 5-year cumulative incidence of symptomatic VTE in the enoxaparin arm (3.8 and 3.6% for THR and TKR, respectively) was similar to that estimated by Lundkvist et al. (3.9%).[6]

In comparison with other studies reflecting Swedish clinical practice,[5,6] the present study followed a similar method by linking cost savings associated with venous thromboembolic event prevention for the cost–effectiveness calculation. Moreover, this study extended the measurement of effectiveness by incorporating QALYs and, therefore, explored the benefit of treatment on patient QoL. In doing so, however, several limitations are attached to the analysis.

Utility scores in the analysis were derived from secondary sources. Ideally, all utility data would be derived from one source to ensure consistency about the applied utility parameters. However, no such source was identified, thus the utility data used in the model were based on several sources. Furthermore, several assumptions were required in order to incorporate these utility scores into the analysis and adjust for the duration of events or age of the model cohort.

Overall, the result of the incremental utility between the two comparators was minimal. This was mainly due to the small disutility associated with venous thromboembolic events and the duration of such episodes within a patient's life-year. The incorporation of utility estimates into the analysis is considered standard practice in the economic evaluation of healthcare interventions, and by doing so the decision-maker is facilitated in comparing investment projects across varying therapeutic areas.

The main limitation of the analysis was the extrapolation of asymptomatic to symptomatic events. An assumption was necessary here to project the difference between the comparator treatments in 'delayed' symptomatic venous thromboembolic events – that is, asymptomatic events becoming symptomatic following the patient evaluation period. This relationship between asymptomatic and symptomatic events is uncertain because tests for asymptomatic VTE are not normally part of routine practice. Furthermore, if asymptomatic VTE is detected, it is always treated; hence, clinical trial follow-up does not reflect what happens in routine practice where VTE is normally undetected and hence untreated. The model assumption allowed us to estimate the potential impact of VTE-related morbidity and infer the health-related QoL impairment associated with each comparator. Sensitivity analysis around this assumption showed that the model results were robust for these parameter changes.

Moreover, in order to complete the extrapolation to 5 years, the analysis bases the risk of long-term complications on an observational study[19] and not the clinical trial. This adds two limitations to the analysis, the first deriving from the heterogeneity of the two studies linked together, and the second by ignoring differences in the diagnosis and treatment of VTE with current practice in Sweden.

To test all the aforementioned structural assumptions and input, we ran extensive sensitivity analyses. The sensitivity analyses did not identify any areas of substantial discrepancy in the model results when the relevant parameters were changed.

Additional research is required on the natural history of VTE and its long-term complications. Nevertheless, it is anticipated that while this additional research would improve the accuracy of the model estimates, there would be little difference in the direction of the results.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.