Article Text

FRI0369 Health Utilities and Costs Among Early Spondyloarthritis Patients Treated and not Treated with TNFα Blockers: Estimates from the Desir Cohort
  1. S. Harvard1,2,3,
  2. A. Molto4,5,
  3. D. Guh3,
  4. N. Bansback2,3,
  5. M. Dougados4,5,6,
  6. S. Paternotte4,7,
  7. P. Richette8,9,
  8. B. Fautrel1,10,
  9. A. Anis2,3
  1. 1UPMC Univ Paris 06, GRC08, Paris, France
  2. 2University of British Columbia
  3. 3Centre for Health Evaluation and Outcome Sciences, Vancouver, Canada
  4. 4Rheumatology Dept., Cochin Hospital
  5. 5Faculty of Medicine, Université Paris Descartes 5
  6. 6EULAR Centre of Excellence
  7. 7Université Paris Descartes 5
  8. 8Rheumatology Dept., Lariboisière Hospital
  9. 9Faculty of Medicine, Université Paris Diderot 7
  10. 10Rheumatology, La Pitié Salpétrière Hospital, Paris, France


Background We previously described the clinical effectiveness of TNFα blockers in a population of matched pairs within the DESIR cohort of early spondyloarthritis (SpA) patients.1 Health utilities and costs have not yet been compared among these patients.

Objectives To compare costs and SF6D health utilities among SpA patients treated and not treated with TNFα blockers in the DESIR cohort.

Methods The DESIR cohort includes 708 patients with early (<3 years) SpA symptoms who are followed-up every 6 months. We examined 202 matched pairs (404 patients) comprised of ‘active’ and ‘controlș patients with/without TNFα blocker use over two years of follow-up. Matching between active and control patients was done by propensity score, with baseline data used to estimate probability of TNFα blocker use. Patient-reported costs from health resource use and productivity loss were valued using French public health and national salary data. SF6D utilities were estimated from SF36 (Version 1) data. Costs and utilities were described only for matched pairs with complete post-baseline cost data. MCMC multiple imputation was used to impute missing SF36 data using BASDAI, BASFI, HAQ, and Ankylosing Spondylitis Quality of Life questionnaire scores. Pairs were compared on baseline variables associated with TNFα blocker use (age, disease duration, BASDAI, BASFI, CRP, ASDAS CRP, NSAID use (yes/no), mNY criteria, MRI sacroilitis, and HLA-B27) using Student's t-tests or Chi-square tests.

Results The analysis included 324 patients. Active and control patients were significantly different on mean BASDAI (active=54.14 vs. control=49.22, p.<0.007), BASFI (active=40.29 vs. control=33.61, p.<0.005), CRP (active=12.73 vs. control=8.89, p.<0.001), ASDAS CRP (active=3.19 vs. control=2.87, p.<0.001), and binary NSAID use (active=0.91 vs. control=0.96, p.<0.001). Mean costs across all follow-ups were € 7062 in the active group and € 1963 in the control group; mean SF6D utility across all follow-ups was in the 0.61113 active group and in 0.63374 the control group.

Conclusions DESIR patients with/without TNFα blocker use were significantly different at baseline in variables known to be associated with costs and quality of life, including BASDAI and BASFI scores. As a result, unadjusted health utilities and costs cannot be compared between these groups. Future studies, including an ongoing cost-effectiveness analysis, must adjust for differences in disease severity between these patients.


  1. Moltό et. al. Arthritis Rheumatol. 2014 Jul;66(7):1734-44.

Acknowledgements We acknowledge the DESIR Cohort and Scientific Committee

Disclosure of Interest None declared

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.