Published in:
01-12-2013 | Original Paper
The cost-effectiveness of TNF-inhibitors for the treatment of rheumatoid arthritis in Swedish clinical practice
Authors:
Ingrid Lekander, Fredrik Borgström, Jörgen Lysholm, Ronald F. van Vollenhoven, Staffan Lindblad, Pierre Geborek, Gisela Kobelt
Published in:
The European Journal of Health Economics
|
Issue 6/2013
Login to get access
Abstract
Objective
The objective was to estimate the cost-effectiveness of TNF-inhibitors for the treatment of rheumatoid arthritis in Swedish clinical practice, both as a first and second biological treatment, with or without the combination of conventional DMARDs. Further sub-group analysis of etanercept treatment was performed.
Methods and materials
Patient level data were obtained from three regions of the Swedish Rheumatology Registers. The dataset contained 2,558 patients who had started TNF-inhibitor treatment, 1,049 with etanercept as their first biological treatment. A total of 819 patients had switched to a second TNF-inhibitor, of which 425 to etanercept. A Markov cohort model was used in which health states of disease severity were classified according to HAQ and DAS28. Disease progression and discontinuation rates of TNF-inhibitors were based on the registry and for the comparator on published literature. Mortality, costs and utilities were based on Swedish data. The main analysis had a societal perspective over 20 years and efficacy was measured in quality-adjusted life-years (QALYs).
Results
TNF-inhibitor treatment was associated with an increase in QALYs and an incremental cost compared to no biological treatment. The cost per QALY gained with the three TNF-inhibitors ranged from €50,000 to €120,000, with lower estimates for TNF-inhibitors used in combination with MTX and as a first biologic. At a progression of 0.045 for the comparator, most values remain within the accepted range for cost-effectiveness.
Conclusions
These results demonstrate that the cost per QALY for TNF-inhibitors was higher than in previous assessments based on registry data and that the results were sensitive to the HAQ progression of the comparator.