Axial spondyloarthritis (ax-SpA) is a chronic inflammatory disorder of the axial skeleton that causes pain, fatigue, stiffness, loss of physical function and impaired health-related quality of life (HRQOL) [1
]. HRQOL is a subjective and multidimensional concept that can be defined as an individual’s experience of their general health state, including their physical, social and mental well-being [5
]. HRQOL can be assessed using generic instruments (e.g. 15D and SF-6D) [6
] that are used in economic evaluation (cost–benefit analysis) and to calculate quality-adjusted life years (QALYs) [6
], and by other generic HRQOL measures such as SF-36 [8
Previous studies have shown that patients with ax-SpA report HRQOL scores that are similar to those reported for other inflammatory diseases [2
], but lower than those in healthy controls [1
]. Women with the ax-SpA report lower HRQOL than men [14
], and decreased HRQOL in patients with ax-SpA is associated with fatigue [17
], increased disease activity, decreased daily activity and exercise [4
], pain, and adverse psychological factors such as body-image disturbance, anxiety and depression [21
]. In the new millennium, effective new treatments have become available for Ax-SpA [e.g. tumour necrosis factor (TNF) inhibitors] [23
] and new treatment strategies have been recommended (the treat-to-target strategy) [23
Only a few studies have examined long-term changes in HRQOL in ax-SpA patients and explored predictors of changes in HRQOL [22
]. Two previous studies found no long-term changes in HRQOL and no major deterioration in clinical outcomes [25
]. The aims of our study were to explore changes in HRQOL over 5 years in patients with ax-SpA and to identify baseline predictors associated with changes in HRQOL assessed using three different measures.
Analysis of the 5-year changes in HRQOL in patients with ax-SpA identified significantly better scores for two of the four main HRQOL measures, SF-6D and SF-36 PCS, while despite increased co-morbidities there were no significant changes in SF-36 MCS and 15D. Baseline characteristics associated with higher 5-year changes in HRQOL across the measures were lower age, higher education, lower BASDAI and BAS-G, and for SF-36 PCS only, no use of biological treatments. The patients seem to have better disease control at the 5-year follow-up compared with baseline, indicated by better scores for measures reflecting disease activity such as CRP, MASES, BAS-G, HAQ and SF-36 bodily pain. One likely explanation for this is that more patients were treated with biologic DMARDs at the 5-year follow-up than at baseline. The increase in use of biologic DMARDs in our study is in line with the results from the Norwegian NOR-DMARD registry. In this registry in the period 2002–2011, the prescription rates for biologic DMARDs increased and disease activity improved in ax-SpA patient [40
]. Further, in the 5-year period of our study the treat to target, recommendations also included ax-SpA [24
To our knowledge, this is the first study since the introduction of the treat-to-target approach that assesses changes in HRQOL in patients with ax-SpA over a 5-year period using three different measures. In contrast to previous studies [22
], we identified better long-term HRQOL within both SF-6D and SF-36 PCS and no changes in SF-36 MCS and 15D. Although the patients had increased co-morbidities at the 5-year follow-up, which we would have expected to influence HRQOL, this did not seem to be the case. The treat-to-target approach might also partly explain the better HRQOL scores [23
]. Better disease control implies better functionality, mobility and less structural damage, all of which have been shown previously to be associated with HRQOL [44
The differences in changes over 5 years using different generic HRQOL measures might be attributable to the nature of the health items included and the way the questions were asked [6
]. However, most generic instruments intended for HRQOL assessment, including SF-6D, 15D and SF-36 as used in the present study, contain at least some items that focus upon physical, emotional and social functioning [47
]. The significant changes in SF-36 PCS and SF-6D might also be attributable to a better responsiveness compared with the other instruments [43
]. However, the effect sizes of the changes were rather weak, and it could be argued that these differences are of limited clinical relevance [43
Baseline self-reported outcome measures reflecting disease activity and burden, together with younger age and higher education, were significant characteristics associated with better long-term HRQOL. This indicates that a lower disease burden at baseline was important for long-term HRQOL. Stable disease and low disease burden have also been identified previously as associated with stable HRQOL over time [22
]. The new treatment-to-target approach implies that medications are adapted to an individual’s disease activity and hence influence their HRQOL [48
]. This is underlined by better SF-36 bodily pain and physical function scores at the 5-year follow-up.
A higher education level was also associated with better long-term HRQOL. Some previous studies have identified associations between education level and HRQOL [22
], but consistent with our findings Ward et al. [49
] found a positive association between higher education level and high HRQOL in patients with ax-SpA. With respect to the effect of age, younger patients have had the disease for a shorter time and are thus likely to be less affected by years of disease and structural damage. Disease duration may also be difficult to calculate in patients with ax-SpA as patients may have had symptoms years before final diagnosis is established, e.g. in ankylosing spondylitis, a phenotype of ax-SpA, it may take 5–10 years before structural changes, a result of inflammation, become visible on X-rays [50
]. With the introduction of magnetic resonance imaging, the inflammatory disease process in the sacroiliac joints in ax-SpA can be visualised years before structural changes become visible [51
]. Further, there may also be a referral delay from general practitioners to rheumatologist that again may delay the diagnosis of ax-SpA.
Although most of the factors significantly associated with better HRQOL assessed by different instruments were the same, we also identified predictors associated with individual measures, such as the association of no use of biological DMARDs with SF-36 PCS only. This might also be attributable to the different items included in this part of the measure, its response options and also the structures of the scores [8
]. Other studies have also identified an association between biological treatment and HRQOL using SF-36 [48
]. On the other hand, our findings could be explained by the limited need for such treatment in patients with low disease activity.
Despite an increased number of co-morbidities over the 5-year period, there seemed to be a trend among ax-SpA patients towards improving disease control, as reflected by lower disease activity, better self-reported functioning and less pain, and a healthier lifestyle as indicated by fewer smokers. The improvement in disease activity in our ax-SpA patients during follow-up may also be explained by a treat-to-target approach and health professional close collaboration with each patient, including encouragement to a healthier lifestyle and a more targeted medication [24
A strength of this study is its long-term design. Another strength is the use of three different HRQOL assessment tools. Further, the study included a relatively large number of patients followed over a 5-year period, and many variables (objective measures and generic and disease-specific patient-reported outcome measures) were included both at baseline and at the 5-year follow-up.
Weaknesses of the study are that data were collected at only two time points. Although we identified changes, the timing of these changes is unknown or unclear, and another two or three time points for data collection would have made the results clearer. This would have been especially beneficial when it comes to prior use of NSAID, synthetic or biological DMARDs in patients not on treatment at follow-up. The 140 patients who were not invited to 5-year follow-up or the panel attrition might have contributed different results. Previous studies show that panel attrition tend to be less healthy than patients taken part [52
]. However, there were only minor baseline differences between those who attended for follow-up and those who were lost to follow-up. A given time for stopping the invitations to 5-year follow-up at one of the hospitals also increased the possibility for random panel attrition. Multivariable linear regression analysis, backward procedure, might produce an overfitted model. However, the explanatory variables chosen for the first step of the model include what can be considered as standard variables, like demographic variables (e.g. gender and age), and variables based on previous studies and theory within the field, along with strong associations in the univariate analyses, and we choose to include a relative high number of clinical variable (which also the high number of patients allows for). Furthermore, the results may be biased by unobserved heterogeneity which could be a threat to the validity of the findings. Adding the lagged dependent variable (HRQOL baseline score) as explanatory variable in panel data may cause biases [53
]. However, the HRQOL baseline score is a clinical and theoretical important variable to predict the clinical development of patients. The time period from baseline to follow-up was long (5 years). We therefore choose to include it in the statistical analysis due to its clinical relevance and predictive ability.
It can be considered as both a strength and a limitation that self-reported outcome measures reflecting disease activity and burden were strongly associated with changes in HRQOL measured by generic questionnaires. While it seems that the disease-specific measures captured the actual disease burden, some of the items included in both the disease-specific and generic questionnaires are similar and therefore might be strongly correlated.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.