Article Text
PDF
Abstract
Objective To determine modifiable factors associated with poor quality of life (QoL) in patients with axial spondyloarthritis (axSpA).
Methods Analysis of data from the British Society for Rheumatology Biologics Register in Ankylosing Spondylitis (BSRBR-AS) and validation of a previous model using data from 1810 patients with axSpA recruited during 2012–2017. Data collected included clinical and patient-reported measures. QoL was assessed using the Ankylosing Spondylitis Quality of Life (ASQoL) measure. Linear regression models predicting ASQoL scores were used first to validate a previous model from a national study, to extend this with additional information available in BSRBR-AS and finally to identify a ‘de novo’ model from BSRBR-AS of which factors impact on poor QoL.
Results Four out of five factors included in a previous model of poor QoL in patients with axSpA were confirmed: Bath Ankylosing Spondylitis Disease Activity Index and Bath Ankylosing Spondylitis Functional Index, fatigue and widespread pain, although the performance of the model was improved by the addition of measures of mood and sleep disturbance. In a de novo model in BSRBR-AS, there were six factors (other than disease activity and function) that predicted ASQoL: depression (β=0.16), sleep disturbance (β=0.08), activity impairment (β=0.04), fibromyalgia (Symptom Severity Scale (β=0.24) and Widespread Pain Index (β=0.10)) and tobacco smoking (β=0.66).
Conclusion This study confirms that poor QoL in patients with axSpA, in addition to high disease activity and poor function, is independently influenced by sleep disturbance, mood and widespread pain. These additional factors are not considered targets for treatment in current European League Against Rheumatism (EULAR) guidelines for managing the condition.
- quality of life, axial spondyloarthritis
- disease register
- fatigue
- mood
- sleep
Statistics from Altmetric.com
Key messages
What is already known about this subject?
Maximising quality of life is the stated aim of EULAR management recommendations in axSpA, and the focus is on reducing disease activity and improving function.
What does this study add?
Poor mental health, sleep problems, and widespread pain all additionally, independently, contribute to poor quality of life in patients with axSpA.
How might this impact on clinical practice or future developments?
Management of axSpA needs to focus on a wider set of targets than disease activity and function.
Introduction
When treating people with axial spondyloarthritis (axSpA) and other inflammatory conditions, rheumatologists are focused on reducing disease activity and, in so doing, aim to reduce the impact which the disease has on their lives. The ultimate aim, however, is to improve patients’ quality of life (QoL). Reducing disease activity is one way to do that, but there may be other factors not directly captured by disease activity measures, which impact on QoL. axSpA has an impact on people’s working lives; mental health and physical health symptoms, such as pain and fatigue, have been shown to have an important influence on QoL.1 2 Further, we (and others) have shown that pharmacological therapy targeted at reducing disease activity in inflammatory arthritis may have modest effects on aspects such as mental health,3 fatigue4 and work productivity.5
Previously, in an analysis of 959 patients from a national disease register (Scotland Registry for Ankylosing Spondylitis (SIRAS)), we have shown that five potentially modifiable factors predict poor QoL (using the Ankylosing Spondylitis Quality of Life Scale (ASQoL)6): high disease activity, poor physical function, fatigue, chronic widespread pain (CWP) and poor spinal mobility.7 Of these factors, disease activity had the lowest (20%) population attributable risk for poor QoL. In addition, there were a number of nonmodifiable factors or at least not easily modifiable in the clinic, which were also related to poor QoL: female sex, fewer years of education, not in full-time employment, living in areas with higher deprivation, not being able to drive and history of peripheral joint involvement. We concluded that ‘these findings provide evidence that in addition to traditional clinical targets…, focus on nonspecific symptoms (CWP and fatigue), perhaps with nonpharmacological therapies, may yield important improvements in QoL’. The positive predictive value for poor QoL varied from around 0% and 15% in those with one or two modifiable factors to around 60% and 80% in those with four or five such factors, respectively.
The aims of the current study, using a Great Britain-wide registry of axSpA (British Society for Rheumatology Biologics Register in Ankylosing Spondylitis (BSRBR-AS)), were (1) to validate the previous model of modifiable factors linked to poor QoL; (2) given that the BSRBR-AS has collected a wider set of variables, to determine whether these additional variables (related to mood and sleep disturbance) are independent in predicting poor QoL; and (3) using the BSRBR-AS to develop a model predicting poor QoL ‘de novo’, to determine how consistent are the factors that predict poor QoL across both populations.
Methods
The BSRBR-AS is a prospective cohort study of people with axSpA. Patients were naïve to biological therapy on recruitment, but some were about to start such therapy (the biological cohort), while others continued on other therapy (non-biological cohort). The study protocol has been published previously.8 Briefly, recruitment took place across 83 secondary care centres between December 2012 and December 2017, initially for those patients, aged at least 16 years, meeting the Assessment of Spondyloarthritis International Society (ASAS) imaging criteria for axSpA9 or the modified New York (mNY) definition of ankylosing spondylitis.10 From November 2014, those meeting the ASAS clinical criteria were also eligible. Clinical data were collected from medical notes, and patients completed questionnaires that were handed out in the clinic and could be completed there, or at home and posted back to the recruitment centre. For the current study, the data used were from the time of recruitment (for non-biological cohort) and just prior to commencing biological therapy (biological cohort), which was also mainly at the time of recruitment. QoL was assessed by ASQoL,6 an 18-item questionnaire which gives a score between 0 (best QoL) and 18 (worst QoL).
Clinical information included extraspinal manifestations (uveitis, psoriasis, inflammatory bowel disease, enthesitis and dactylitis), inflammatory markers (C reactive protein), peripheral joint involvement, symptom duration, body mass index (BMI) and information on 14 comorbidities (related to cardiovascular, respiratory, gastrointestinal, renal and neurological conditions and cancer). Patient-reported measures included age, gender, level of education, employment status and lifestyle factors (tobacco smoking and alcohol intake), as well as Bath Ankylosing Spondylitis Disease Activity Index (BASDAI),11 Bath Ankylosing Spondylitis Functional Index (BASFI)12 and Bath Ankylosing Spondylitis Metrology Index (BASMI).13 These Bath indices all produce scores ranging from 0 to 10 (least to most severe). Participant postcodes were used to determine a deprivation quintile, with reference to either the population of Scotland or England and Wales.14–16 This ranged from 1 (most deprived) to 5 (least deprived). Mental health was assessed by the Hospital Anxiety and Depression Scale (HADS) scored from 0 (best) to 21 (worst).17 Overall work impairment (including absenteeism, presenteeism) and other activity (non-work) impairment by the Work Productivity and Activity Impairment (WPAI) Specific Health Problem, all scored from 0% to 100%.18 19 The Widespread Pain Index (WPI) (0–19) and Symptom Severity Scale (SSS) scores were assessed through the 2011 fibromyalgia ‘research’ criteria.20 This was collected only among persons recruited from August 2015. Fatigue was collected through the Chalder Fatigue Scale (0–33)21 and sleep disturbance by the Jenkins Sleep Evaluation Questionnaire (0–20),22 with higher scores on each indicating worse state.
Patients attended meetings to identify priority areas for analysis.
Analysis
In validating the model linking modifiable factors to QoL previously reported by Dean et al,7 the five modifiable factors reported (fatigue, BASFI, CWP, BASDAI and BASMI) were included in a multivariable linear regression model. The model was adjusted for non-modifiable factors associated with QoL on univariable analysis at p≤0.2. The current study did not have information available to determine the presence of CWP according to the American College of Rheumatology 1990 criteria for fibromyalgia23 and instead used the WPI subscale of the 2011 ‘research’ criteria for fibromyalgia.20 Building on this five-factor model, the second analysis offered three additional modifiable factors (anxiety, depression and sleep disturbance), which were available in BSRBR-AS but had not been available within SIRAS. The stepwise selection therefore had eight candidate variables. Variables were entered into the model at p≤0.1 and excluded at p≥0.15. Adjustment for nonmodifiable factors was applied as previously. The third analysis was a multivariable linear regression model, de novo, with forward stepwise selection, using modifiable factors from BSRBR-AS with p≤0.2 from the univariable analysis. The model examined which factors, in addition to disease activity and function, influenced poor QoL but omitted the work productivity factors (absenteeism, presenteeism and work impairment), which were only available for employed participants. Adjustment for nonmodifiable factors was applied as previously. For each model, once the variables to be included were determined, the model was re-run using all the participants with data for the included variables (rather than only participants with data available for each of the candidate variables).
All statistical analysis was undertaken using STATA V.15 and on the August 2017 version of the BSRBR-AS dataset.
Results
A total of 1810 participants were eligible for the current analyses. Approximately two-thirds were male (67%), their median age was 49 years (IQR 38–61), with a median time since symptom onset of 17 years (IQR 8–31) (table 1). Of those who had been tested, 80% were HLA-B27 positive. Most participants (67%) met the mNY criteria for ankylosing spondylitis (AS), an additional 29% fulfilled ASAS imaging criteria but not mNY, and 4% fulfilled only ASAS clinical criteria for axSpA. The median BASDAI and ASQoL scores were 4.8 (IQR 2.5–6.8) and 9 (IQR 3–13), respectively.
Characteristics of the British Society for Rheumatology Biologics Register in Ankylosing Spondylitis patients
Factors associated with poor QoL
Among clinical factors, all extraspinal manifestations, with the exception of uveitis, were associated with poorer QoL (table 2). Higher BMI and a greater number of comorbidities were also associated with poorer QoL. Longer symptom duration, however, was associated with better QoL (β per year=−0.05, 95% CI −0.07 to −0.04). Most patient-reported factors demonstrated a relationship. Worse disease activity, function and metrology were significantly associated with poorer QoL (BASDAI β per unit increase=1.82, BASFI β=1.56, BASMI β=0.97). Females reported slightly poorer QoL (β=1.58, 95% CI 1.02 to 2.13), and there were important associations with low socioeconomic status (eg, those who had attended only secondary school/those living in the highest levels of deprivation had on average around four points higher ASQoL score than those with a further degree/living in an area with lowest levels of deprivation, respectively). Current smoking was associated with significantly worse QoL (β=4.45, 95% CI 3.74 to 5.16). Higher level of fatigue (β per one unit increase in score=0.63, 95% CI 0.60 to 0.67), fibromyalgia (SSS) (β=1.34 per unit increase, 95% CI 1.24 to 1.44), fibromyalgia (WPI) (β=0.69, 95% CI 0.61 to 0.77), sleep disturbance (β=0.56, 95% CI 0.52 to 0.59), anxiety (β=0.72, 95% CI 0.67 to 0.76) and depression (β=0.97, 95% CI 0.92 to 1.01) were all related to poorer QoL. Non-work activity impairment due to AS (β=0.16 per % impairment, 95% CI 0.15 to 0.16) was also associated with poor QoL. Among those working, a higher percentage of time absent (β=0.11 per % time absent, 95% CI 0.09 to 0.13) or working with reduced productivity (presenteeism) (β=0.14 per % time work impaired, 95% CI 0.13 to 0.15) were related to poor QoL.
Predictors of Ankylosing Spondylitis Quality of Life from univariable linear regression analysis
Validation of model predicting poor QoL
There were 555 participants in the current study who provided the necessary information to be part of the validation of the QoL model reported in the SIRAS study (the lower number was due mainly to the fact that WPI was only collected partway through the study, and there were missing data for BASMI). Four of the previously reported five factors (BASDAI (β=0.69, 95% CI 0.51 to 0.87), BASFI (β=0.85, 95% CI 0.69 to 1.01), fatigue (β=0.14, 95% CI 0.08 to 0.19) and widespread pain (WPI) (β=0.07, 95% CI 0.00 to 0.14)) were related to poor QoL, but there was no independent relationship with BASMI (β=0.01, 95% CI −0.16 to 0.18) (table 3). A stepwise model was then run with these five factors and additional psychosocial factors available in the current study. All three extra added factors (sleep disturbance (β=0.10, 95% CI 0.07 to 0.12), anxiety (β=0.12, 95% CI 0.09 to 0.16) and depression (β=0.19, 95% CI 0.14 to 0.24)) were included in the new best-fitting model, together with disease activity (BASDAI) (β=0.55, 95% CI 0.45 to 0.64) and function (BASFI) (β=0.85, 95% CI 0.77 to 0.93) (table 4).
Validation of a model predicting poor quality of life
Validation of a model predicting poor QoL: the additional role of depression, anxiety and sleep disturbance
Developing a new model predicting poor QoL
Finally, a de novo stepwise model within BSRBR-AS used data from 642 participants. The patient-reported factors included in the final model were disease activity (BASDAI (β=0.31, 95% CI 0.14 to 0.47)), function (BASFI (β=0.59, 95% CI 0.45 to 0.73)), depression (β=0.16, 95% CI 0.09 to 0.24) and sleep disturbance (β=0.08, 95% CI 0.04 to 0.13), in agreement with results of previous models (table 5). In addition, activity impairment (β=0.04, 95% CI 0.02 to 0.05), fibromyalgia (SSS) (β=0.24, 95% CI 0.13 to 0.35) and fibromyalgia (WPI) (β=0.10, 95% CI 0.03 to 0.17) entered the model as did current tobacco smoking (β=0.66, 95% CI 0.10 to 1.21).
Variables associated with poor quality of life: the British Society for Rheumatology Biologics Register in Ankylosing Spondylitis study
Discussion
The results of this study provide consistent evidence of which modifiable factors are associated with poor QoL in patients with axSpA. A model previously proposed in a Scotland-wide study was partly validated, confirming that worse disease activity, poor function and high levels of fatigue and widespread nature of pain symptoms were strongly related to poor QoL. The results were then extended by showing that in addition, mood (anxiety and depression), sleep disturbance, and both widespread pain and somatic symptom components related to fibromyalgia were importantly related to poor QoL.
The participants in BSRBR-AS are broadly representative of the patients with axSpA in clinics across Great Britain, with the exception that none of them were currently prescribed biologics (although around one-third of the participants were about to commence biologics). We have used BASDAI as the measure of disease activity as it is the measure most commonly used in the UK and is part of national guidelines by the National Institute of Health and Clinical Excellence and the British Society for Rheumatology. A second reason is that the necessity to have a measure of inflammation (such as for the Ankylosing Spondylitis Disease Activity Score)24 would have resulted in a much higher level of missing data, where this had not been measured in the required timeframe.
Some models have many fewer subjects included than others. This was partly a result of variables related to fibromyalgia (WPI and SSS), which were only collected for part of the duration of the study and in relation to BASMI, which was only measured on some participants. Once the variables included in a specific model were determined, the analysis was, however, re-run, including all participants with the required data available. In validating the model predicting poor QoL, the information collected was not in exactly the same format between studies (for pain, WPI was collected instead of CWP). However, the information collected in BSRBR-AS was more detailed and was analysed in a statistically more efficient manner (ie, using continuous variables where available). Overall the results, however, were consistent across both studies. There can be a certain amount of circularity, since factors that are potentially associated with poor QoL can themselves be used in assessing QoL. Within ASQoL, for example, there are items on function, mental health, sleep and pain. These are all aspects (assessed by specific questionnaires) that the current study has found were associated with overall QoL. Such circularity is unavoidable, but in the current study, the regressions model the association per unit change in score, so there is limited influence of one aspect of QoL on the overall score and second in our analysis as part of the SIRAS study, items were removed (in turn) from ASQoL on pain and tiredness, and associations with CWP and fatigue (respectively) were still observed.7 Finally, one may debate whether function is modifiable independent of disease activity. In this dataset, there is a clear correlation between them (correlation=0.76). We have considered both as EULAR recommendations include each as targets for management, and the results suggest they make an independent contribution to QoL. The correlations between SSS and BASDAI (correlation=0.68) and between BASFI and non-work physical impairment as measured by WPAI (correlation=0.78) also suggest important relationships. However, as part of the model assessment, we calculated the variance inflation factor as a method of assessing potential multicollinearity, and this confirmed that there were no concerns.
The current study confirms the important role of disease activity and function in terms of QoL but adds to the literature by emphasising the important independent role of additional features associated with axSpA: mental health, fatigue and sleep problems, and widespread pain. Fatigue has been recognised to have an important influence on health-related QoL in rheumatoid arthritis (RA)25 and indeed it was a priority for RA patients who participated in a focus group study in Sweden, in terms of being a key component to measure in evaluating treatment success.26 Zhao et al,27 in a meta-analysis of 16 studies, estimated the prevalence of at least moderate depression in patients with axSpA as 15%, based on a HADS score of ≥11. Garrido-Cumbrera et al,28 in a sample of 680 patients as part of the Atlas of Axial Spondyloarthritis in Spain, reported that high disease activity was a risk factor for poor mental health, but the current study emphasises that poor mental health independently predicts worse QoL. A meta-analysis of the co-occurrence of fibromyalgia in axSpA estimated a prevalence of 13%,29 compared with a prevalence in the general population of around 2%–5%,30 and we have previously shown within BSRBR-AS that patients who have comorbid fibromyalgia have the same absolute improvement in QoL when treated with anti-TNFα therapy, but their QoL prior to and on treatment remains worse. Further, a high score on the SSS (rather than the WPI) is predicative of lack of improvement.31 The fact that these additional features are common has been recognised, but not that they contribute independently to poor QoL, and there has been a lack of studies on how they can be effectively treated (including alongside inflammatory arthritis). Nevertheless, we acknowledge that there are some who propose that it is sufficient to focus on inflammation and that, in so doing, other aspects that impinge on quality of life will also improve.32
The most recent EULAR/ASAS guidelines for the management of axSpA state, in Recommendation 2, that ‘the primary goal of treating the patient with axSpA is to maximise long-term health-related quality of life through control of symptoms and inflammation, prevention of progressive structural damage, preservation/normalisation of function and social participation’.33 The results of the current study confirm that disease activity and function, as a focus, are appropriate. However, it is not sufficient. In the guidelines, there is no mention of sleep problems, widespread pain or mental health and specifically how these aspects should be managed. The results from this study suggest that their role is important, and independent of disease activity and functional limitation. Results from others studies, as noted previously, suggest that patients will continue to experience fatigue, poor mental health and fibromyalgia-like symptoms if management is focused on inflammation alone.3 4 Effecting improvement in such additional disease features is challenging; studies are under way to test behavioural approaches to management and/or physical activity for fatigue and fibromyalgia symptoms in patients with a range of inflammatory arthritides.34 35 Currently, evidence suggests, for example, that community-based exercise programmes exert a positive (although modest) effect on anxiety36 37 and depression38 among patients with arthritis and other rheumatic conditions. A recent trial demonstrated that group-based cognitive behavioural therapy delivered within rheumatology teams reduced the impact of fatigue in patients with RA.39 Such therapy aims to reduce the impact of, for example, fatigue and widespread pain rather than improving symptoms per se, and not all patients are willing to engage with them. Further, the expertise and resources to deliver the interventions to target these additional factors are not easily available to many rheumatology teams.
In summary, the current study has shown that, through analysis of factors related to poor QoL and validation of a previously published model, improving the QoL of patients with axSpA means that, in addition to improving disease activity and function in patients, there must be attention to the comorbid features of fatigue, poor sleep and mental health and other common symptoms.
Acknowledgments
We are grateful to the staff who contributed to running the British Society for Rheumatology Biologics Register in Ankylosing Spondylitis register and to the recruiting staff at the clinical centres, details of which are available online (https://www.abdn.ac.uk/iahs/content-images/181128BSRBR-ASFullContributorList.pdf)
References
Footnotes
Handling editor Josef S Smolen
Twitter @UAberdeenEpi, @hteraG_senoJ
GJM and OR contributed equally.
Contributors All authors discussed and contributed to designing this study. GJM, OR and LED designed the analysis plan, which was undertaken by OR and overseen by LED and GJM. Results were reviewed by all authors. OR and GJM both contributed to drafting the first version of the manuscript, which was critically reviewed by all authors.
Funding The British Society for Rheumatology Biologics Register in Ankylosing Spondylitis is funded by the British Society for Rheumatology, which has received funding for this from Pfizer, AbbVie and UCB. These companies receive advance copies of manuscripts for comments but made none in relation to this manuscript.
Disclaimer Pfizer, AbbVie and UCB have no input in determining the topics for analysis or work involved in undertaking it.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The study was approved by the National Research Ethics Service Committee North East-County Durham and Tees Valley (Research Ethics Committee reference 11/NE/0374), and informed consent was obtained from all participants.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data may be obtained from a third party and are not publicly available.