Fatigue independently predicts different work disability dimensions in etanercept-treated rheumatoid arthritis and ankylosing spondylitis patients

Inflammatory rheumatic diseases, such as rheumatoid arthritis (RA) and ankylosing spondylitis (AS), are a substantial cause of work disability [1], a key outcome for both the patient and society. Estimates indicate that within 10 years of symptom onset up to 70% of patients become work disabled [2], rising to approximately 90% as disease duration increases [3].

Biological therapies have significantly minimised the physical sequelae of poorly controlled inflammation, such as joint destruction [4–6]. This in turn has been associated with an improvement in patient outcomes including those relating to work capacities [4–8]. Despite these gains, work disability remains an issue for the majority of RA and AS patients and it has become increasingly clear that the determinants of work disability in inflammatory diseases are not necessarily inflammatory in origin [9–11].

Fatigue, a patient priority inconsistently associated with inflammation [12], is possibly one such determinant. This symptom represents a persistent, unmanageable, and unpredictable issue [13–15] which is reported by up to 80% of patients with AS and RA. The impact of fatigue is multi-dimensional. For example, levels of physical activity are reduced and relationships are harmed [13, 14]. Yet the symptom is poorly managed, and many patients feel this subject is ignored by their clinicians. Subsequent attempts to self-manage by trial and error are typically unsuccessful [14, 16].

Anecdotally, patients have identified fatigue as a principal barrier to employment and productivity [13, 16]. The subsequent need to reduce work hours, relinquish roles, or even retire from work altogether have been attributed to the emotional, physical, and cognitive consequences of the symptom [3, 13, 16, 17]. However, the empirical evidence supporting these attributions is limited.

To date, analyses of the relationship between fatigue and work disability have generally examined cross-sectional data in small sample sizes [3, 18] and knowledge regarding the long-term nature of the relationship is sparse. Indeed, a previous systematic review of work disability predictors in RA surprisingly failed to identify any studies that have evaluated fatigue as a putative factor [19]. Furthermore, no study designs examining the impact of fatigue have employed a comprehensive measure of work disability which is necessary to adequately capture this complex multi-dimensional construct. To evaluate the different components, which include presenteeism (reduced performance at work due to ill health) and productivity in addition to the traditionally applied measure of absenteeism, suitably designed and detailed tools should be applied [20]. Finally, to our knowledge, no investigation has compared the fatigue-work disability relationship between distinct rheumatic disorders.

We have previously observed that patients commonly report continuing high levels of fatigue despite successfully attaining clinical remission (Disease Activity Score (DAS28) < 2.6) 6 months after commencing biologic therapy [21]. We now hypothesise that, in this current era of biological therapies, fatigue is independently associated with continuing levels of work disability among patients with inflammatory rheumatic diseases. Specifically, this study aimed to examine whether fatigue independently predicts different work disability dimensions, adjusting for clinical and psychosocial factors, over 1 year in working patient populations commencing etanercept, a commonly prescribed biologic. Furthermore, we obtained data from two near-identical studies on AS and RA to compare the relative impact of fatigue on work disability between these two common, but pathologically distinct, inflammatory rheumatic diseases.

Data were obtained from 6518 patients recruited to one of two observational studies on patients with clinician-confirmed AS (n = 1663; clinical study number NCT00544557) or RA (n = 4855; clinical study number NCT00488475) who were commencing etanercept as part of routine clinical practice in Germany. Ethical approval for the original studies was obtained from the Ethics Committee of the State Chamber of Physicians (Landesaertzekammer; NCT00544557, 21 September 2007; NCT00488475, 15 September 2006).

Both cohorts were followed-up for 1 year. Data collected at baseline, 6 months, and 1 year contributed to this study. Of the variables captured, data obtained for these analyses comprised fatigue, work disability, demographics, clinical, and psychosocial status.

Apart from the expected greater proportion of male AS versus RA patients, the cohorts were comparable at baseline. No relevant differences in clinical, psychosocial, or work disability domains were observed.

In this large study, fatigue had a significant and independent effect on all work disability dimensions at 6 and 12 months for patients with RA commencing etanercept. For AS patients who commenced etanercept, fatigue was identified as having a significant effect on presenteeism and overall activity impairment only. Across both studies, the lack of significant interactions between fatigue and visit suggests a consistent effect of baseline fatigue over the following 12-month period.

This study adds to the growing body of evidence highlighting the great importance of this previously ignored symptom among patients with inflammatory rheumatic diseases, and particularly emphasizes the economic impact of fatigue for both the individual and society as a whole. The existence of this relationship in the context of expensive biological treatment serves to enhance the economic issues further—despite their prescription, work disability remains an issue. The data presented herein support the case for targeting fatigue in parallel (employing, for example, non-pharmacological approaches) with a view to optimizing work-related outcomes and so providing even greater societal returns to the considerable pharmacological investments that global health systems have made in recent years. Furthermore, the study uniquely indicates that fatigue is importantly associated with work disability across two disparate disease populations, although the apparent differential effect on work disability dimensions should encourage some consideration of disease-specific issues during the development of appropriate interventions.

The few previous studies to quantitatively examine the relationship between RA and AS fatigue and work disability have been mostly cross-sectional and have all reported a significant association between these two factors [3, 18, 26, 27]. Of the longitudinal assessments, Zirkzee and colleagues found fatigue to be only weakly associated (p = .131) with deteriorating work status at 1 year among a small, likely underpowered, sample (n = 69) of Dutch patients newly diagnosed with inflammatory arthritis [28]. Similarly, fatigue was identified as a significant univariable, but not multivariable, predictor of receiving a work disability pension in a Norwegian study (n = 159) that aimed long-term work disability (the follow-up time point was 7 years) [29]. However, for patients, the predictors of shorter-term work disability are likely to be more informative in the context of preventative intervention development. Such future therapies may rescue patients from more immediate personal financial concerns related to their employment.

Pre-existing deficiencies in work disability measures were consistently the strongest predictors of the same outcome at 1 year. While this is intuitive, more surprising was the absence of independent associations between disease activity or pain and work disability.

This observation may reflect the success of biological therapies in targeting these factors during the follow-up of these cohorts and subsequently attenuating any association. Although fatigue may also be a mediator of the relationship between pain or disease activity and work disability, it is also plausible that no strong relationship may actually exist between these variables. Certainly, a previous systematic review only reported inconsistent evidence for a relationship between these traditional clinical factors [19].

In keeping with a previous report [30], we have shown that levels of fatigue and work disability are comparable between RA and AS working age populations. However, in this study, fatigue appears to differentially impact the domains of work impairment in RA and AS. One possible explanation for this difference may be the (expected) higher prevalence of males within the AS cohort. Work disability has previously been found to be dependent on the work status of the spouse [9], i.e. those married to a working partner are more likely to reduce/stop their own work when compromised by disease. In many cultures, including in Germany, it remains common for males to adopt the traditional role as the principal economic provider for the family unit. Thus, the threshold for work disability in response to fatigue could be higher as a result of their greater financial responsibilities. Another explanation is that the impact of fatigue may be greater for females compared with males. In rheumatic disease, fatigue is consistently reported to be greater among females for reasons as yet undetermined [31]. Finally, the nature of fatigue (and subsequent impact) may be distinct between AS and RA resulting from separate pathways.

Certainly, in terms of developing preventative interventions, the differential impact of fatigue as well as the more prominent association between anxiety/depression and work disability for AS compared with RA patients supports the delivery of disease-specific rather than generic interventions to prevent work disability.

A number of potential limitations should be considered when interpreting these results. Firstly, participants were new commencers of biological therapies and so were not representative of all AS and RA patients. The majority of RA and AS patients do not receive biological therapies [32, 33]. Such patients are typically characterised by high levels of disease activity and past disease-modifying anti-rheumatic drug failure. Data from both the German and UK biologics registers, which also recruit disease controls, indicate that RA patients not receiving biologicals are demographically similar, but AS patients tend to be younger [34–36]. Although it is possible that these results may not be generalisable, the population frame (which enables the selection of those patients with significant work disability despite a trial of biological therapy) is ideal to help us elucidate those determinants of work disability that are not necessarily fully related to inflammation (such as fatigue). We also acknowledge that generalisability of our identified associations may be influenced by cultural differences. For example, attitudes towards work [37], resilience factors [38], and availability of social security [26] support may influence the levels of work disability and all may vary between countries (although this may not necessarily confound the relationship with fatigue).

Secondly, in the absence of a control group, we cannot quantify the effect of etanercept on either fatigue or work productivity. Nevertheless, additional analyses showed that, among those still taking etanercept at 12 months (RA n = 1091, AS n = 689), average reductions of 20% were seen in presenteeism, work productivity loss, and activity impairment in both cohorts while absenteeism reduced by an average of 8–15% in RA and AS, respectively (data not shown).

Furthermore, work impairment is likely multifactorial in origin. While we were able to access data on a number of factors which are commonly implicated with this outcome, a number of other potentially interesting variables, for example sleep disturbance, co-medications, and disease severity, were not captured by the study and could not be incorporated into this analysis. Thirdly, given the observational design of this study, causality cannot be inferred although it seems quite possible given what we already know from patient reports. Finally, the psychosocial measures available to us for this secondary analysis provided only non-specific global measures of the underlying constructs of interest. For example, the EQ-5D does not enable the relative contributions of anxiety and depression to be separately determined.

Similarly, we acknowledge that our use of a VAS to capture fatigue may not have fully captured this complex construct relative to multi-dimensional measures, such as the Bristol Rheumatoid Arthritis Fatigue Multi-Dimensional Questionnaire (BRAF MDQ) or Chalder Fatigue Questionnaire (CFQ) [39]. Fatigue VAS scales are, however, considered suitable for use for those seeking to obtain a global measurement of fatigue, and are sensitive to change and are commonly used, aiding comparison with other diseases [39].

Future studies should seek to validate these finding using different fatigue instruments in different sampling frames. The development of interventions to reduce work disability among patients with arthritis is a key priority for patients and researchers alike. The targets for such interventions are likely to be multiple, but this study provides evidence that fatigue-specific measures should be a key component in the development of any future interventions. Existing evidence supports the efficacy of non-pharmacological interventions such as exercise and cognitive behavioural therapies for fatigue in the context of inflammatory rheumatic diseases [40]. A better understanding of the biology of fatigue will be important for the development of alternative therapies in the future.

This is the largest study to examine the longitudinal association between fatigue and work impairment and, by selecting two cohorts which underwent near-identical interventions and data collection procedures, the first to enable the comparison of this relationship between two different rheumatic diseases. We have demonstrated that, among patients beginning etanercept therapy, fatigue has a significant effect on the absenteeism, presenteeism, productivity loss, and activity impairment of patients with RA, and a significant but dimension-selective effect on work disability among AS patients. In addition to pharmacological treatment of inflammation, fatigue-alleviating interventions such as cognitive behavioural therapy and exercise should be tested in both RA and AS patients with a view to modifying the course of this crucial outcome.