Jaana T Joensuu1, Kalle J Aaltonen2, Pasi Aronen2, Tuulikki Sokka3, Kari Puolakka4, Riitta Tuompo5, Markku Korpela6, Mikko Vasala7, Kirsti Ilva8, Dan Nordström9, Marja Blom2. 1. Faculty of Pharmacy, University of Helsinki, Helsinki jaana.joensuu@helsinki.fi. 2. Faculty of Pharmacy, University of Helsinki, Helsinki. 3. Department of Rheumatology, Jyväskylä Central Hospital, Jyväskylä 4. Department of Medicine, South Karelia Central Hospital, Lappeenranta. 5. Department of Medicine, Helsinki University Central Hospital, Helsinki. 6. Centre for Rheumatic Diseases, Tampere University Hospital, Tampere. 7. Department of Internal Medicine, Kainuu Central Hospital, Kajaani. 8. Department of Medicine, Kanta-Häme Central Hospital, Hämeenlinna. 9. Department of Medicine, Helsinki University Central Hospital, Helsinki Faculty of Medicine, University of Helsinki, Helsinki, Finland.
Abstract
OBJECTIVE: The aim of this study was to explore the cost-effectiveness of biological DMARDs (bDMARDs) compared with conventional synthetic DMARDs (csDMARDs) for RA using real-world data from Finnish registers. METHODS: RA patients starting their first bDMARD and comparator patients using csDMARDs during 2007-11 were obtained from the National register of biologic treatments in Finland and the Jyväskylä Central Hospital patient records. Propensity score matching was applied to adjust for differences between bDMARD and csDMARD users. Effectiveness was measured in quality-adjusted life years (QALY) and based on the register of biologic treatments in Finland and Jyväskylä Central Hospital patient records, whereas the direct costs were obtained from relevant Finnish national registers. Patients were followed up for 2 years, and both costs and effectiveness for the second year were discounted at 3%. The incremental cost-effectiveness ratio (ICER) with 95% CI was calculated based on bootstrapped mean costs and effectiveness. RESULTS: Of 1581 RA patients meeting study inclusion criteria, 552 bDMARD and 220 csDMARD users were included in analyses after matching. Mean costs for bDMARDs and csDMARDs were €55 371 and €24 879, while mean effectiveness was 1.23 and 1.20 QALYs, respectively. Consequent ICER was €902 210/QALY. Results were confirmed in sensitivity analyses. CONCLUSION: The high incremental cost and the small, non-significant difference in effectiveness resulted in high ICER, suggesting that bDMARDs are not cost-effective. Regardless of matching, latent confounders may introduce bias to the results.
OBJECTIVE: The aim of this study was to explore the cost-effectiveness of biological DMARDs (bDMARDs) compared with conventional synthetic DMARDs (csDMARDs) for RA using real-world data from Finnish registers. METHODS: RA patients starting their first bDMARD and comparator patients using csDMARDs during 2007-11 were obtained from the National register of biologic treatments in Finland and the Jyväskylä Central Hospital patient records. Propensity score matching was applied to adjust for differences between bDMARD and csDMARD users. Effectiveness was measured in quality-adjusted life years (QALY) and based on the register of biologic treatments in Finland and Jyväskylä Central Hospital patient records, whereas the direct costs were obtained from relevant Finnish national registers. Patients were followed up for 2 years, and both costs and effectiveness for the second year were discounted at 3%. The incremental cost-effectiveness ratio (ICER) with 95% CI was calculated based on bootstrapped mean costs and effectiveness. RESULTS: Of 1581 RA patients meeting study inclusion criteria, 552 bDMARD and 220 csDMARD users were included in analyses after matching. Mean costs for bDMARDs and csDMARDs were €55 371 and €24 879, while mean effectiveness was 1.23 and 1.20 QALYs, respectively. Consequent ICER was €902 210/QALY. Results were confirmed in sensitivity analyses. CONCLUSION: The high incremental cost and the small, non-significant difference in effectiveness resulted in high ICER, suggesting that bDMARDs are not cost-effective. Regardless of matching, latent confounders may introduce bias to the results.