Article Text

This article has a correction. Please see:

Extended report
Double-blind placebo-controlled trial of etanercept in the prevention of work disability in ankylosing spondylitis
  1. Nick Barkham,
  2. Laura C Coates,
  3. Helen Keen,
  4. Elizabeth Hensor,
  5. Alexander Fraser,
  6. Anthony Redmond,
  7. Lorna Cawkwell,
  8. Paul Emery
  1. Academic Unit of Musculoskeletal Disease, University of Leeds, Leeds, UK
  1. Correspondence to Professor Paul Emery, Academic Unit of Musculoskeletal Disease, University of Leeds, Chapel Allerton Hospital, Chapeltown Road, Leeds LS7 4SA, UK; p.emery{at}


Objectives Etanercept has been shown to be rapidly effective in suppressing disease activity in ankylosing spondylitis (AS). The aim of this study was to determine whether etanercept improves work instability as measured by the Ankylosing Spondylitis Work Instability Scale (AS-WIS).

Method Forty patients with active AS who were in work but were work unstable were recruited. Patients were randomised to receive 25 mg etanercept or placebo twice weekly for 12 weeks. The primary outcome was change in AS-WIS at week 12. The AS-WIS is a patient-derived outcome measure which allows stratification of the risk of job loss. Secondary outcomes included clinical outcomes and gait parameters.

Results The mean improvement in AS-WIS score at week 12 was 2.75 in the etanercept group and 0.68 in the placebo group (p=0.125). The risk of job loss decreased for 11 (55%) of the etanercept group compared with 7 (35%) in the placebo group. Conversely, the risk of job loss increased in 3 (15%) of the placebo group compared with 1 (5%) in the etanercept group. There was no statistically significant difference between treatment groups in change in WIS categories (Mann–Whitney U test=0.153, p=0.160). Significant improvement with etanercept was seen at week 12 in clinical outcomes and gait parameters. Etanercept was well tolerated, with no dropouts due to adverse events.

Conclusion This small study confirms the efficacy of etanercept on clinical outcome measures in patients with AS and suggests an effect on work instability which needs to be replicated in a larger controlled study.

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.

The onset of ankylosing spondylitis (AS) is usually among those of working age (in the third decade or even earlier).1 The impact of AS on employment status is significant. In a Dutch study, overall participation in the labour force was 54.2% for the AS cohort, a significant reduction of 11% compared with the general population of the same age.2 More than 75% of patients with AS who had stopped working were officially recognised as work disabled.

A Swedish study using prospectively collected registry data to investigate sick leave in patients with AS compared with population-based controls matched for age, sex and residential area found that more patients with AS than controls were registered for sickness benefit and sickness compensation.3 Cases had an increased risk of sick leave compared with controls, with a relative risk of 1.8 and a median of 30 additional sick days per year than controls. AS carries a significant economic burden arising from both the direct costs of medical and disability care and from the indirect costs associated with loss of earnings and reduced productivity.

The advent of tumour necrosis factor α (TNFα) blockade marks the first therapeutic advance in AS since the introduction of non-steroidal anti-inflammatory drugs (NSAIDs). A randomised double-blind placebo-controlled study of the safety and efficacy of etanercept in AS showed significant improvement in the etanercept group in axial symptoms, function and quality of life.4

Work instability is defined as a state in which the consequences of a mismatch between an individual's functional abilities and the demands of his or her job can threaten continuing employment if not resolved.5 Work instability represents the patient perspective as it reflects the difficulties in meeting job requirements. On the other hand, it has the potential to be a surrogate for work production, an economically relevant outcome. Ideally, an instrument measuring this concept should be sensitive to change and have a good predictive validity.

The aim of our study was to determine whether etanercept treatment improves work instability as measured by the Ankylosing Spondylitis Work Instability Scale (AS-WIS),6 and to assess the clinical outcomes associated with change in working ability.


The trial was conducted at the Leeds Teaching Hospitals Trust, Leeds, UK. Patients were recruited consecutively. All had a definite diagnosis of AS by modified New York Criteria7 and active disease as defined by two out of three of Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) ≥40 (0–100), visual analogue scale (VAS) pain ≥40 (0–100), early morning stiffness ≥45 min. All were in work but were work unstable (AS-WIS score >10). Exclusion criteria included past or current tuberculosis, congestive heart disease or treatment with glucocorticoids in the previous month.

Treatment protocol

Patients were randomised in a 1 to 1 ratio to receive subcutaneous etanercept (25 mg) or placebo twice weekly for 12 weeks (ie, 24 doses.) The site monitor, investigators and patients remained blinded until after the data through week 12 had been finalised. The patients were permitted to remain on a stable dose of the disease-modifying antirheumatic drugs (DMARDs) sulfasalazine or methotrexate and/or an NSAID for the duration of the study. Patients were not permitted corticosteroids or other DMARDs.

Study assessments

The primary outcome in the study was change in the work instability of patients after 12 weeks measured using the AS-WIS.6 The AS-WIS is a self-administered 20-item questionnaire which allows stratification of those experiencing a mismatch between their functional capacity and the demands of their job into categories of risk for job loss. The scale was generated from qualitative interviews with patients with AS combined with items from the Rheumatoid Arthritis Work Instability Scale.5 The scale was validated against a gold standard of expert workplace assessment, which allowed AS-WIS score cut-off points representing low (<11), medium (11–18) and high (19–20) risk of job loss to be determined.6

Secondary outcomes in this study included assessments of disease activity (BASDAI),8 quality of life (Ankylosing Spondylitis Quality of Life (ASQoL) instrument),9 functional ability (Bath Ankylosing Spondylitis Functional Index (BASFI)),10 gait parameters using an electronic walkway and disability (Disability Index of Stanford Health Assessment Questionnaire (HAQ-DI)).11 Work participation was assessed prospectively using work diaries. Hours lost due to AS were expressed as a proportion of the number of hours that the patient would usually have worked.

Sample size

In the absence of existing data on the effects of etanercept on work instability in patients with AS, we assumed we might see a similar magnitude of effect to that seen in other outcomes in previous studies.4 12 13 We assumed a reduction in AS-WIS score of 4 points in patients receiving etanercept compared with 2 points in the placebo group (SD 2.0), with 80% power at the 5% level of significance, which required 20 patients to be recruited to each treatment group allowing for 15% dropout.

Statistical analysis

Analysis was performed on an intention-to-treat basis, with the last available value imputed in the case of missing data. Two-tailed tests were used throughout and 95% CIs are provided where appropriate. Analysis of covariance (ANCOVA) was used to compare changes in work instability, disease activity, functional impairment and gait parameters at 12 weeks between the treatment groups, taking baseline values as covariates. To facilitate parametric analysis, the ordinal AS-WIS data were Rasch-transformed14 to interval scaling prior to ANCOVA. Changes in risk of job loss defined by cut-off points for (untransformed) AS-WIS were compared between groups using the Mann–Whitney U test. Proportions of patients losing working hours were compared using χ2 analysis. Ankylosing Spondylitis Assessment Group (ASAS40)response was calculated (≥40% and ≥2 points improvement in three of four domains (patient global VAS, pain VAS, BASFI, last two questions of the BASDAI) with no worsening at all in the remaining domain), in addition to BASDAI50 (≥50% improvement in BASDAI). To assess the responsiveness of the AS-WIS, standardised response means (SRM: mean change/SD of the change) were calculated for ASAS40/BASDAI50 responders and non-responders. Pearson's correlations (r) were used to explore relationships between objective functional measures and subjective outcomes reported using ASQoL and AS-WIS. Spearman's rho correlations were used to explore the relationship between baseline AS-WIS score and the proportion of working hours lost due to AS during follow-up. All analyses were performed using SPSS Version 16.0.2.


A total of 52 patients were screened. Forty were enrolled in the study, 20 of whom were randomised to receive active treatment. Twelve patients were excluded after screening. Three patients were lost to follow-up between screening and randomisation, three did not meet radiographic criteria for AS and six did not have sufficient disease activity or work instability to meet the inclusion criteria. The baseline characteristics of the patients are shown in table 1.

Table 1

Demographic data and baseline clinical, gait and occupational results for each treatment group

Primary outcome

When considering the AS-WIS at the group level, there was a mean (95% CI) improvement in absolute AS-WIS score at week 12 (adjusted for baseline values) of −2.75 (−4.63 to −0.87) in the etanercept group versus −0.68 (−2.56 to 1.21) in the placebo group, a difference which did not reach statistical significance (F=2.47, p=0.125). Analysis of ‘completers only’ did not greatly affect significance (p=0.109).

When analysed at the individual level, change in risk category for job loss was considered. Risk of job loss decreased for 11 (55%) of the etanercept group (eight from medium to low risk, three from high to medium risk) compared with seven in the placebo group (35%). Conversely, risk of job loss increased in three (15%) of the placebo group compared with one (5%) in the etanercept group. There was no statistically significant difference between treatment groups in terms of change in WIS categories (Mann–Whitney U=153, p=0.160; see table 1 in online supplement).

Secondary outcomes

There were significant improvements (p<0.05) in the etanercept group for BASDAI, BASFI, ASQoL, HAQ-DI and gait outcomes (see table 1 in online supplement). For results by job type, working hours lost and relationship between primary and secondary outcomes please see text and table 2 in online supplement. Etanercept was well tolerated with no serious adverse events and no dropouts due to adverse events (for further details of adverse events see online supplement).


This is the first study of anti-TNF therapy in AS to use work instability as the primary outcome measure, using a work instability scale which has been specifically validated in patients with AS.

The study relies on the AS-WIS scale, which has some limitations. The scale has only recently been developed from a small sample of patients and it has not undergone external validation. The predictive validity towards job retention is unknown. Sensitivity to change has yet to be fully assessed; in the current study the standardised response mean for the eight patients who achieved a BASDAI50 response was 1.65 compared with 0.17 for the 32 patients who did not achieve BASDAI50. This finding indicates that the AS-WIS is highly responsive, but it needs to be replicated in a larger group of patients. Ultimately, the risk categories in the AS-WIS scale are defined on the basis of expert opinion.

The improvement in work instability seen in the etanercept group in this study did not achieve statistical significance. However, the cohort of patients had a long disease duration and the study may have been too short to demonstrate a significant improvement in work instability, which may lag behind clinical improvement as people take time to adjust their working practices in line with improvement in functionality. Certainly, the study was too short to show an effect on withdrawal from work; longer term follow-up is required to confirm that the improvements in work instability are sustained and do indeed result in job retention.

If the improvements seen in disease activity, objective functional measures and work instability in patients treated with etanercept do translate into job retention, this suggests that the use of expensive therapies may be justified on health economic grounds.


The authors would like to thank Mike Horton and Professor Alan Tennant from the LIMM Academic Department of Rehabilitation Medicine for their assistance in the Rasch transformation of the AS-WIS.


View Abstract

Supplementary materials

  • Web Only Data

    Files in this Data Supplement:

  • Web Only Data

    Files in this Data Supplement:


  • Funding Financial support for the study, and the study drug was provided by Wyeth Pharmaceuticals. Wyeth had no role in the study design, or in the collection, analysis, or interpretation of the data.

  • Competing interests None.

  • Ethics approval The local ethics committee approved the study and written informed consent was obtained from all patients prior to screening.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • Miscellaneous
    BMJ Publishing Group Ltd and European League Against Rheumatism