Article Text

OP0065 Quantifying the Economic Impact of Serious Infections from Anti-TNFS for Rheumatoid Arthritis: Results from the BSRBR-RA
  1. E. M. Heather1,2,
  2. K. Payne1,
  3. M. J. Harrison1,
  4. B. Control Centre Consortium2,
  5. K. L. Hyrich on behalf of the British Society for Rheumatology Biologics Registers2,
  6. D. P. Symmons2,3
  1. 1Manchester Centre for Health Economics
  2. 2Arthritis Research UK Epidemiology Unit, Centre for Musculoskeletal Research, Manchester Academic Health Science Centre, The University of Manchester
  3. 3National Institute for Health Research Manchester Musculoskeletal Biomedical Research Unit, Manchester, United Kingdom


Background Anti-tumour necrosis factor drugs (anti-TNFs) for adult rheumatoid arthritis (RA) are associated with an increased risk of serious infections (SIs) compared to non-biologic disease modifying drugs (nbDMARDs). SIs, requiring hospitalisation or outpatient treatment with IV antibiotics, will likely impact on the relative cost-effectiveness (CE) of anti-TNFs given the associated increased use of scarce healthcare resources.

Objectives To examine whether the length of hospital stay (LoS) following a SI could be predicted on the basis of treatment received and baseline patient characteristics.

Methods Patients with RA treated with nbDMARDs or anti-TNFs (adalimumab; etanercept; infliximab) who experienced a SI in the first year of follow-up were identified from the BSRBR-RA database. Patient characteristics recorded at baseline included: age; sex; disease duration; treatment history; comorbidities; HAQ; and DAS28. Data on SIs and related LoS were collated from (i) bi-annual patient diaries (ii) bi-annual rheumatologist questionnaires (iii) death certificate data from the NHS Information Centre. SIs were ascribed to the most recent drug recorded, if diagnosed whilst on-drug or within 90 days of the first missed dose. Patients were (i) censored after a first SI (ii) excluded if switching anti-TNFs within the first year. Kruskal-Wallis tests were used for between-treatment comparisons. A generalised linear regression model (log-link function; negative-binomial distribution) was fitted to identify the key factors driving LoS in hospital following a SI. Potential confounders were included as covariates and missing baseline data imputed.

Results Of the 15395 patients contributing 14027 patient years (pyrs) of follow-up (nbDMARDs: 3295 pyrs; anti-TNF: 10732 pyrs), 838 experienced a SI in the first year of treatment. Unadjusted rates of SIs were significantly higher in the anti-TNF cohort: 67.5 (95% CI 62.6-72.5) vs. 35.2 (28.5-41.6) per 1000 pyrs (p<0.001). 83 nbDMARD and 590 anti-TNF patients were hospitalised. The mean LoS did not differ significantly between the nbDMARD and anti-TNF cohort (11.3 vs. 10.7 days; p=0.46). The key factors associated with an increased LoS were: female gender (+2.3 days); age ≥75years (+4.7 days); having 1-3 comorbidities (+2.0 days); failing 1-5 nbDMARDs (+2.3 days). There was no significant difference in the LoS between anti-TNFs.

Conclusions Previous findings - that SIs occur more frequently with anti-TNFs compared to nbDMARDs - are confirmed. LoS was related to demographic factors, prior treatment history and co-morbidity but not to current treatment. Nevertheless anti-TNFs appear to result in additional healthcare costs from SI-related hospitalisations due to their increased frequency. These findings are important for understanding the relative economic impact of anti-TNFs and indicate that SIs should be considered when considering the CE of these agents.

Acknowledgements EH is supported by an NIHR Methods Fellowship.

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.