You are here

Article Text

Article menu

Download PDFPDF

Clinical and epidemiological research
Extended report
Disease activity trajectories in early axial spondyloarthritis: results from the DESIR cohort
  1. Anna Molto1,2,3,
  2. Sophie Tezenas du Montcel4,5,
  3. Daniel Wendling6,
  4. Maxime Dougados2,3,
  5. Antoine Vanier4,7,
  6. Laure Gossec1,8
  1. 1Institut Pierre Louis d'Epidémiologie et de Santé Publique, Sorbonne Universités, UPMC Univ Paris 06, GRC-08, Paris, France
  2. 2Rheumatology Department, Paris Descartes University, Cochin Hospital, AP-HP, Paris, France
  3. 3Clinical Epidemiology and Biostatistics, PRES Sorbonne Paris-Cité, INSERM (U1153), Paris, France
  4. 4Department of Biostatistics Public Health and Medical Informatics, UPMC Université Paris 06, AP-HP, Pitié Salpêtrière Hospital, Paris, France
  5. 5Sorbonne University, Université Pierre et Marie Curie (UPMC) Univ Paris 6, UMR_S 1136, Institut Pierre Louis d'Epidémiologie et de Santé Publique, Paris, France
  6. 6Rheumatology Department, CHRU de Besançon, Université de Franche-Comté, Besançon, France
  7. 7EA 4275 SPHERE, University of Nantes, Nantes, France
  8. 8Department of Rheumatology, AP-HP, Pitié Salpêtrière Hospital, Paris, France
  1. Correspondence to Dr Anna Molto, Rheumatology B Department, Cochin Hospital, 27 rue du Faubourg Saint Jacques, Paris 75014, France; anna.molto{at}aphp.fr

Abstract

Background Disease activity may change over time in axial spondyloarthritis (axSpA). The objectives were to identify patterns of disease activity evolution in patients with early axSpA.

Methods Patients from the prospective early axSpA cohort (DEvenir des Spondyloarthrites Indifférenciées Récentes (DESIR)) who fulfilled the Assessment in SpondyloArthritis Society (ASAS) criteria for axSpA at baseline and with at least three Ankylosing Spondylitis Disease Activity Score (ASDAS) values available over the 3 years of follow-up were analysed. Statistical analyses: trajectories were estimated by group-based trajectory modelling; predisposing baseline factors for such trajectories were identified by univariate and multivariable multinomial (logit) regression; work disability over time was compared between the trajectories by Cox hazard model.

Results In all, 370 patients were analysed: mean disease duration was 1.6 (±0.9) years. The five distinct trajectories of disease activity over the 3 years were (t1) ‘persistent moderate disease activity’ (n=134 (36.2%)); (t2) ‘persistent inactive disease’ (n=66 (17.8%); (t3) ‘changing from very high disease activity to inactive disease’ ((n=29 (7.8%)); (t4) ‘persistent high disease activity’ (n=126 (34.1%)) and (t5) ‘persistent very high disease activity’ (n=15 (4.1%)). After adjustment for other characteristics, t2 was associated with a white-collar job (OR=2.6 (95% CI 1.0 to 6.7)) and t3 with male gender (OR=7.1 (1.6 to 32.2)), higher education level (OR=9.4 (1.4 to 63.4)) and peripheral joint involvement (OR=6.2 (1.23 to 31.32)). Patients from (t4) and (t5) were more often declared work disabled over follow-up (HR=5.2 (1.5 to 18.0) and HR=8.0 (1.3 to 47.9), respectively).

Conclusions Trajectory modelling of disease activity was feasible in early axSpA: more than 30% patients (141/370) were in a trajectory with a persistent high disease activity. Persistent high disease activity trajectories were significantly associated with consequences on work.

Trial registration number NCT01648907.

  • Spondyloarthritis
  • Outcomes research
  • Disease Activity
  • Epidemiology

https://doi.org/10.1136/annrheumdis-2016-209785

Statistics from Altmetric.com

    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.

    Axial spondyloarthritis (axSpA) presentation can be phenotypically heterogeneous, for example, predominant axial involvement, predominant articular peripheral involvement with or without psoriasis, exclusive axial involvement with radiographic sacroiliitis, etc.1 Due to this presentation diversity and also due to the different treatment modalities and other elements (eg, socioeconomic environment and access to healthcare, gender, etc.), disease activity can be heterogeneous, both at presentation and over time.2 ,3 However, studies aiming to identify patterns of disease activity evolution over time in axSpA are sparse.4 Furthermore, a disability is often associated with higher disease activity scores.5 Therefore, a better identification and characterisation of homogeneous groups of patients based on disease activity would allow a better and tailored strategy for the follow-up of patients with axSpA.

    In other disciplines, trajectory modelling has been applied to identify patterns of behaviour,6–8 but only very few studies have applied these methodologies in the field of rheumatic diseases, to identify homogeneous groups of patients over follow-up.9 ,10 To the best of our knowledge, those methods have not been applied in axSpA.

    The Ankylosing Spondylitis Disease Activity Score (ASDAS) based on C-reactive protein (CRP) is a widely validated tool to measure disease activity in axSpA that integrates both patient-reported items and objective inflammatory markers.11 ,12 Given its face validity and its psychometric properties,13 ,14 we proposed to use ASDAS to define the trajectories of disease activity over time in an axSpA cohort.

    Using the DESIR cohort15 data, we aimed to identify (a) disease activity trajectories in patients with early axSpA over a 3-year follow-up period, (b) the baseline characteristics associated with such trajectories and (c) the outcomes associated with each trajectory in terms of treatment and disability.

    Patients and methods

    Study design

    DESIR is a French prospective, multicentre, longitudinal observational cohort aiming to study patients with early inflammatory back pain (IBP) suggestive of SpA (clinicaltrials.gov NCT01648907).15 ,16 This study fulfilled current good clinical practices and has obtained the approval of the appropriate ethical committee. Participants in the study gave their written informed consent.17

    Patients

    A total of 708 patients were included in DESIR: consecutive patients aged >18 and <50 years with IBP according to the Calin et al18 or Berlin19 criteria for more than 3 months but less than 3 years and symptoms suggestive of diagnosis for SpA score ≥5 (on a Numerical Rating Scale of 0–10, where 0=not suggestive and 10=very suggestive of SpA). None were taking tumour necrosis factor α inhibitors (TNFi) at baseline. For our analysis, only patients fulfilling the ASAS criteria for classification of axSpA20 ,21 at baseline and for whom at least three ASDAS values were available during follow-up were included. The data set used was locked in December 2013.

    Collected data

    The collected data comprised demographics and clinical presentation of the disease at baseline and each 6 months for the first 2 years, and at year 3. Demographics included age and gender. Medico-economic data were also collected: highest degree of education; type of employment: blue collar (ie, physically demanding jobs, eg, farmer) versus white collar (ie, sedentary job, eg, secretary); employment state (ie, currently working, on sick leave or in permanent work disability) and days of sick leave over each period.

    Disease activity for this analysis was evaluated by the ASDAS.11 The ASDAS associates several activity criteria: total back pain, peripheral pain/swelling, duration of morning stiffness (questions 2, 3 and 6 of the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI),22 respectively), patient global and CRP, combined in a single parameter. Since methodologies allowing the trajectories definitions require a single variable to define such trajectories,23 this was used to define disease activity. Severity of the disease was assessed by the Bath Ankylosing Spondylitis Functional Index (BASFI).24 Quality of life was evaluated according to the Short Form-36 (SF-36).25 Non-steroidal anti-inflammatory drugs (NSAID) treatment was evaluated by the ASAS-NSAID score26 during the 6 months preceding each study visit. Exposure to TNFi over follow-up was also collected.

    Statistical analysis

    1. Trajectories modelling: To identify distinctive trajectories, we used a semiparametric mixture model: group-based trajectory model (GBTM).23 ,27 GBTMs model the relationship between a variable (here, ASDAS) and time: for each trajectory, the shape of the trajectory and the estimated proportion of the population belonging to each trajectory, and for each patient, the probability to belong to each different trajectory. Each participant is then assigned to the group for which her/his probability to belong to a trajectory is the highest. For this, we used the TRAJ procedure in the SAS software V.9.4: this procedure allows the estimation and comparison of models with several numbers of trajectories and shapes (constant, linear, quadratic or cubic). To select the best-fitting model, models with one to six trajectories and with several shapes were compared. The best-fitting model was selected according to the Bayesian information criterion. It was also required that means of individual posterior probability to belong to a trajectory were all superior to 0.7.

    2. Baseline factors associated with each trajectory: Demographics, disease phenotype, disease severity, quality of life and NSAID treatment at baseline were compared in the different trajectories by univariate multinomial (logit) regression. Thereafter, multivariable multinomial regression was performed to identify the independent baseline factors associated with each trajectory, including in the model the baseline characteristics with a p value <0.10 after the univariate analyses. For these analyses, trajectory 1 (‘persistent moderate disease activity’) was used as the reference trajectory, because it was the trajectory with the largest number of patients and fitted better the data according to the Akaike Index Criterion. ASDAS was not included in the model because it was the parameter defining the trajectory. BASDAI was not included either, because of a concern of colinearity with ASDAS.

    3. Outcomes associated with each trajectory: Exposure to TNFi and disability declaration were collected every 6 months, and evaluated over the 3 years of follow-up as a dichotomous state (‘ever exposed’ and ‘ever disabled’). Both were compared in the different trajectories by Cox hazard models (proportional hazards assumption was confirmed by Schoenfeld residuals). The number of days of sick leave (if patients were declared in permanent work disability, they were considered as in sick leave all year) over follow-up were compared in the different trajectories by linear regression.

    SAS V.9.4 was used for the TRAJ procedure; R software V.3.1.1 was used for the rest of the analyses.

    Results

    Of the 708 patients included in the DESIR cohort at baseline, 439 (62.0%) fulfilled the ASAS criteria for axial SpA at baseline and 370 (52.3%) had at least three ASDAS values available during the 3 years of follow-up (table 1). Among the included patients, the percentage of patients with available data for ASDAS was 359 (82%), 319 (73%), 319 (73%), 295 (67%), 326 (74%) and 219 (66%) for the baseline, 6, 12, 18, 24 and 36 months, respectively.

    View this table:
    Table 1

    Baseline disease characteristics of 370 early patients with axial spondyloarthritis

    Trajectories

    The analyses yielded five distinctive trajectories of disease activity during the 3 years of follow-up (figure 1 and see online supplementary table S1).

    Figure 1
    Figure 1

    Trajectories of disease activity in early axial spondyloarthritis according to the group-based trajectory model technique. ASDAS, Ankylosing Spondylitis Disease Activity Score.

    Trajectory 1 (t1) (n=134 (36.2%)) included patients with ‘persistent moderate disease activity’. Trajectory 2 (t2) (n=66 (17.8%)) included patients with ‘persistent inactive disease’. Trajectory 3 (t3) (n=29 (7.8%)) included patients very high disease activity at baseline but reaching an inactive disease after 12 months and remaining in this state until the 36th month (‘changing disease activity’). Trajectory 4 (t4) (n=126 (34.1%)) included patients who presented with ‘persistent high disease activity’ and trajectory 5 (t5) (n=15 (4.1%)) included patients with ‘persistent very high disease activity’.

    Baseline characteristics associated with each trajectory

    Results of the multivariable analysis are presented in table 2: compared with patients in (t1), patients in (t2) (‘persistent inactive disease’) had more frequently a white-collar job (OR=2.6 (95% CI 1.0 to 6.7)), whereas patients from (t3) (‘changing disease activity’) were more frequently males (OR=7.1 (1.6 to 32.2)) with a higher degree of education (OR=9.4 (1.4 to 63.4) and more frequently a history of peripheral joint involvement (OR=6.2 (1.2 to 31.1)). Poorer quality of life (SF36 mental and physical components) at baseline was significantly associated with high disease activity trajectories in the univariable model, but was not retained in the multivariable model (table 1).

    View this table:
    Table 2

    Characteristics associated with trajectories in early axial spondyloarthritis (multinomial logit regression)*

    (T3) (‘changing disease activity’) and (t5) (‘persistent very high disease activity’) presented with almost identical ASDAS mean values at baseline (4.0 (±0.8) and 4.1 (±0.5) for (t3) and (t5), respectively). We compared these two subgroups in terms of baseline characteristics, and multivariable analysis only evidenced gender (male) (OR=17.59 (2.2 to 424.5)) and university education (OR=12.0 (1.5 to 279.2)) as baseline characteristics independently associated with (t3).

    Outcomes associated with trajectories

    TNFi intake

    Twenty-five (18.7%), 4 (6.1%), 21 (72.4%), 42 (33.3%) and 8 (53.3%) patients received a TNFi during follow-up in (t1), (t2), (t3), (t4) and (t5), respectively (table 3). Compared with (t1), patients from (t3) (‘changing disease activity’) were the group of patients who received more frequently a TNFi over time (HR=4.5 (95% CI 3.5 to 5.9)). Interestingly, patients from trajectories (t4) (‘persistent high disease activity’) and (t5) (‘persistent very high disease activity) also received more TNFi as compared with (t1) (HR=1.8 (1.4 to 2.2) and HR=2.63 (1.8 to 3.9), respectively).

    View this table:
    Table 3

    Outcomes associated with distinct disease activity trajectories in early axial spondyloarthritis

    Work disability

    Patients from trajectories (t1), (t2), (t3), (t4) and (t5) presented a mean (±SD) number of days of sick leave over the 3 years of follow-up of 43 (±127), 15 (±41), 22 (±36), 75 (±116) and 300 (±312), respectively. Patients from (t5) were significantly more frequently on sick leave over follow-up (p<0.001) compared with patients from (t1). Over the 3 years, 1.5%, 3.0%, 0%, 7.9% and 13.3% patients from (t1), (t2), (t3), (t4) and (t5), respectively, were considered work disabled. Patients from (t4) and (t5), the trajectories with persistent high disease activity, were significantly more frequently declared work disabled over time (HR=5.2 (1.5 to 18.0) and HR=8.0 (1.3 to 47.9), for (t4) and (t5), respectively). Interestingly, despite the initial very high disease activity state, no patients from (t3) (‘changing disease activity’) were declared work disabled over follow-up (table 3).

    Discussion

    In the era of personalised medicine and tailored treatment strategies, the identification of disease evolution is important to improve patients' management. Here, we have applied an original and validated methodology to determine longitudinal patterns of disease activity in an early axSpA cohort. This study identified five disease activity trajectories: two trajectories with stable moderate/low disease activity (t1 and t2), two trajectories with stable high/very high disease activity (t4 and t5) and a disease activity improving trajectory (t3); 141/370 patients (38%) belonged to trajectories of persistent disease activity.

    These results highlight that even in a country with wide access to biologics,28 axSpA remains a disease where more than a third of patients could remain in moderate to high disease activity over several years.

    Nevertheless, some baseline characteristics were strongly associated with stable low and improving disease activity trajectories, that is, being a male, a higher degree of education and having a white-collar job. These results are consistent with what has been previously reported in rheumatoid arthritis (RA): in the COMOrbidities in Rheumatoid Arthritis (COMORA) cross-sectional study that included 3920 patients with RA worldwide, after adjustment, women (vs men) and low-educated (vs university) patients had higher disease activity.29 In the field of SpA, feminine gender has also been found to be associated with higher disease activity reported by the BASDAI despite lower acute phase reactants in several clinical trials,30–35 and high-rank occupation has been found to be associated with lower disease activity in patients with SpA.36 It is difficult to determine the causality of such links. Are educated men receiving better treatment (though of note, here, patients belonging to trajectories of persistent low disease activity received less frequently TNFi over follow-up), are they more adherent, do they have less severe disease or are they complaining less? Since ASDAS, the main criterion to define active disease here, is a mixed objective and subjective criterion, it is difficult to conclude on this point. In any case, physicians should be aware that when facing a patient with early axSpA, females with less formal education may be more at risk of persistent disease activity.

    Another finding from this study was related to sick leave and work disability in early axSpA. First, the rate was rather high in this cohort (16/182 patients with available data on work disability, 10%). Second, it was strongly related to the disease activity trajectory, which validates both the methodology used here and the use of ASDAS as an outcome to assess disease activity. Such validations are important in the field of axSpA where assessments are often subjective and have not always been validated in terms of prediction of later outcomes.

    Our study has several limitations and also some strengths. The main strength of our study is the innovative methodology allowing to evaluate disease activity patterns longitudinally. This validated methodology has been used in other disciplines, rarely in rheumatology and never in SpA. Furthermore, the large sample of patients presenting with early axSpA, according to the ASAS classification criteria, has allowed us to define distinctive trajectories of disease activity from an early time after onset of the disease. Nevertheless, most trajectories revealed a stable disease activity over follow-up and the main baseline characteristics associated with trajectories were demographic and socioeconomic. It is not impossible that the subjective patient-reported outcomes included in the ASDAS contributed more to the trajectories definitions rather than CRP. However, it is worth noting that the current guidelines recommend using both patient-reported outcomes and acute phase reactants (eg, CRP) for disease activity monitoring in SpA.37

    Also, one may argue why ASDAS trajectories were not adjusted for TNFi use over time. TNFi use is associated with an important decrease of ASDAS.38 ,39 Therefore, when modelling ASDAS trajectories over time, ASDAS trajectories inherently include TNFi use, and can thus be considered a reflection of the course of disease including its treatment.

    Finally, we only assessed the outcome of the different trajectories in terms of disability and days of sick leave, and not in terms of structural damage (ie, radiographic sacroiliitis or syndesmophyte formation). However, structural progression is known to be very slow in axSpA40 and particularly in this cohort,41 that it did not seem appropriate to use such outcome for the first 3 years of follow-up.

    Further studies evaluating longitudinally disease activity and the long-term outcomes of the different patterns of disease activity are needed to determine the validity of such trajectories in other patient groups.

    References

    1. ↵
      1. Dougados M,
      2. Baeten D
      . Spondyloarthritis. Lancet 2011;377:2127–37. doi:10.1016/S0140-6736(11)60071-8
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Chung HY,
      2. Machado P,
      3. van der Heijde D, et al
      . Smokers in early axial spondyloarthritis have earlier disease onset, more disease activity, inflammation and damage, and poorer function and health-related quality of life: results from the DESIR cohort. Ann Rheum Dis 2012;71:809–16. doi:10.1136/annrheumdis-2011-200180
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Chung HY,
      2. Machado P,
      3. van der Heijde D, et al
      . HLA-B27 positive patients differ from HLA-B27 negative patients in clinical presentation and imaging: results from the DESIR cohort of patients with recent onset axial spondyloarthritis. Ann Rheum Dis 2011;70:1930–6. doi:10.1136/ard.2011.152975
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Paramarta JE,
      2. De Rycke L,
      3. Ambarus CA, et al
      . Undifferentiated spondyloarthritis vs ankylosing spondylitis and psoriatic arthritis: a real-life prospective cohort study of clinical presentation and response to treatment. Rheumatology (Oxford) 2013;52:1873–8. doi:10.1093/rheumatology/ket239
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Boonen A,
      2. Boone C,
      3. Albert A, et al
      . Understanding limitations in at-work productivity in patients with active ankylosing spondylitis: the role of work-related contextual factors. J Rheumatol 2015;42:93–100. doi:10.3899/jrheum.131287
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Blanchard CM,
      2. McSweeney J,
      3. Giacomantonio N, et al
      . Distinct trajectories of light and moderate to vigorous physical activity in heart disease patients: results from the Activity Correlates afTer cardIac hospitalizatiON (ACTION) trial. J Sci Med Sport 2014;17:72–7. doi:10.1016/j.jsams.2013.03.002
      OpenUrl
    7. ↵
      1. Pryor LE,
      2. Tremblay RE,
      3. Boivin M, et al
      . Developmental trajectories of body mass index in early childhood and their risk factors: an 8-year longitudinal study. Arch Pediatr Adolesc Med 2011;165:906–12. doi:10.1001/archpediatrics.2011.153
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Nordon C,
      2. Rouillon F,
      3. Azorin JM, et al
      . Trajectories of antipsychotic response in drug-naive schizophrenia patients: results from the 6-month ESPASS follow-up study. Acta Psychiatr Scand 2014;129:116–25. doi:10.1111/acps.12135
      OpenUrlCrossRefPubMed
    9. ↵
      1. Holla JFM,
      2. van der Leeden M,
      3. Heymans MW, et al
      . Three trajectories of activity limitations in early symptomatic knee osteoarthritis: a 5-year follow-up study. Ann Rheum Dis 2014;73:1369–75. doi:10.1136/annrheumdis-2012-202984
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Siemons L,
      2. Ten Klooster PM,
      3. Vonkeman HE, et al
      . Distinct trajectories of disease activity over the first year in early rheumatoid arthritis patients following a treat-to-target strategy. Arthritis Care Res 2014;66:625–30. doi:10.1002/acr.22175
      OpenUrl
    11. ↵
      1. Lukas C,
      2. Landewé R,
      3. Sieper J, et al
      . Development of an ASAS-endorsed disease activity score (ASDAS) in patients with ankylosing spondylitis. Ann Rheum Dis 2009;68:18–24. doi:10.1136/ard.2008.094870
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Machado P,
      2. Landewé R,
      3. Lie E, et al
      . Ankylosing Spondylitis Disease Activity Score (ASDAS): defining cut-off values for disease activity states and improvement scores. Ann Rheum Dis 2011;70:47–53. doi:10.1136/ard.2010.138594
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. van der Heijde D,
      2. Lie E,
      3. Kvien TK, et al
      . ASDAS, a highly discriminatory ASAS-endorsed disease activity score in patients with ankylosing spondylitis. Ann Rheum Dis 2009;68:1811–18. doi:10.1136/ard.2008.100826
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Fernández-Espartero C,
      2. de Miguel E,
      3. Loza E, et al
      . Validity of the ankylosing spondylitis disease activity score (ASDAS) in patients with early spondyloarthritis from the Esperanza programme. Ann Rheum Dis 2014;73:1350–5. doi:10.1136/annrheumdis-2012-202976
      OpenUrlAbstract/FREE Full Text
    15. ↵
      1. Dougados M,
      2. Etcheto A,
      3. Molto A, et al
      . Clinical presentation of patients suffering from recent onset chronic inflammatory back pain suggestive of spondyloarthritis: The DESIR cohort. Joint Bone Spine 2015;82:345–51. doi:10.1016/j.jbspin.2015.02.006
      OpenUrlCrossRefPubMed
    16. ↵
      1. Dougados M,
      2. d'Agostino MA,
      3. Benessiano J, et al
      . The DESIR cohort: a 10-year follow-up of early inflammatory back pain in France: study design and baseline characteristics of the 708 recruited patients. Joint Bone Spine 2011;78:598–603. doi:10.1016/j.jbspin.2011.01.013
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      La cohorte DESIR. http://www.lacohortedesir.fr/ (accessed 30 Jan 2015).
    18. ↵
      1. Calin A,
      2. Porta J,
      3. Fries JF, et al
      . Clinical history as a screening test for ankylosing spondylitis. JAMA 1977;237:2613–14. doi:10.1001/jama.1977.03270510035017
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Rudwaleit M,
      2. Metter A,
      3. Listing J, et al
      . Inflammatory back pain in ankylosing spondylitis: a reassessment of the clinical history for application as classification and diagnostic criteria. Arthritis Rheum 2006;54:569–78. doi:10.1002/art.21619
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Rudwaleit M,
      2. Landewé R,
      3. van der Heijde D, et al
      . The development of Assessment of SpondyloArthritis international Society classification criteria for axial spondyloarthritis (part I): classification of paper patients by expert opinion including uncertainty appraisal. Ann Rheum Dis 2009;68:770–6. doi:10.1136/ard.2009.108217
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Rudwaleit M,
      2. van der Heijde D,
      3. Landewé R, et al
      . The development of Assessment of SpondyloArthritis international Society classification criteria for axial spondyloarthritis (part II): validation and final selection. Ann Rheum Dis 2009;68:777–83. doi:10.1136/ard.2009.108233
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Garrett S,
      2. Jenkinson T,
      3. Kennedy LG, et al
      . A new approach to defining disease status in ankylosing spondylitis: the Bath Ankylosing Spondylitis Disease Activity Index. J Rheumatol 1994;21:2286–91.
      OpenUrlPubMedWeb of Science
    23. ↵
      1. Nagin DS,
      2. Odgers CL
      . Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol 2010;6:109–38. doi:10.1146/annurev.clinpsy.121208.131413
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Calin A,
      2. Garrett S,
      3. Whitelock H, et al
      . A new approach to defining functional ability in ankylosing spondylitis: the development of the Bath Ankylosing Spondylitis Functional Index. J Rheumatol 1994;21:2281–5.
      OpenUrlPubMedWeb of Science
    25. ↵
      1. Jenkinson C,
      2. Coulter A,
      3. Wright L
      . Short form 36 (SF36) health survey questionnaire: normative data for adults of working age. BMJ 1993;306:1437–40. doi:10.1136/bmj.306.6890.1437
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Dougados M,
      2. Simon P,
      3. Braun J, et al
      . ASAS recommendations for collecting, analysing and reporting NSAID intake in clinical trials/epidemiological studies in axial spondyloarthritis. Ann Rheum Dis 2011;70:249–51. doi:10.1136/ard.2010.133488
      OpenUrlAbstract/FREE Full Text
    27. ↵
      1. Nagin DS,
      2. Odgers CL
      . Group-Based Trajectory Modeling (Nearly) Two Decades Later. J Quant Criminol 2010;26:445–53. doi:10.1007/s10940-010-9113-7
      OpenUrlCrossRefPubMed
    28. ↵
      1. Putrik P,
      2. Ramiro S,
      3. Kvien TK, et al
      . Inequities in access to biologic and synthetic DMARDs across 46 European countries. Ann Rheum Dis 2014;73:198–206. doi:10.1136/annrheumdis-2012-202603
      OpenUrlAbstract/FREE Full Text
    29. ↵
      1. Putrik P,
      2. Ramiro S,
      3. Keszei AP, et al
      . Lower education and living in countries with lower wealth are associated with higher disease activity in rheumatoid arthritis: results from the multinational COMORA study. Ann Rheum Dis 2016;75:540–6. doi:10.1136/annrheumdis-2014-206737
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. van der Horst-Bruinsma IE,
      2. Zack DJ,
      3. Szumski A, et al
      . Female patients with ankylosing spondylitis: analysis of the impact of gender across treatment studies. Ann Rheum Dis 2013;72:1221–4. doi:10.1136/annrheumdis-2012-202431
      OpenUrlAbstract/FREE Full Text
    31. ↵
      Gender-Attributable Differences in Outcome of Ankylosing Spondylitis: Long-Term Results from the Outcome in Ankylosing Spondylitis International Study. ACR Abstr. http://acrabstracts.org/abstracts/gender-attributable-differences-in-outcome-of-ankylosing-spondylitis-long-term-results-from-the-outcome-in-ankylosing-spondylitis-international-study/ (accessed 24 Mar 2015).
    32. ↵
      1. van der Heijde D,
      2. Da Silva JC,
      3. Dougados M, et al
      . Etanercept 50 mg once weekly is as effective as 25 mg twice weekly in patients with ankylosing spondylitis. Ann Rheum Dis 2006;65:1572–7. doi:10.1136/ard.2006.056747
      OpenUrlAbstract/FREE Full Text
    33. ↵
      1. Davis JC,
      2. Van Der Heijde D,
      3. Braun J, et al
      . Recombinant human tumor necrosis factor receptor (etanercept) for treating ankylosing spondylitis: a randomized, controlled trial. Arthritis Rheum 2003;48:3230–6. doi:10.1002/art.11325
      OpenUrlCrossRefPubMedWeb of Science
    34. ↵
      1. Calin A,
      2. Dijkmans BA,
      3. Emery P, et al
      . Outcomes of a multicentre randomised clinical trial of etanercept to treat ankylosing spondylitis. Ann Rheum Dis 2004;63:1594–600. doi:10.1136/ard.2004.020875
      OpenUrlAbstract/FREE Full Text
    35. ↵
      1. Braun J,
      2. van der Horst-Bruinsma IE,
      3. Huang F, et al
      . Clinical efficacy and safety of etanercept versus sulfasalazine in patients with ankylosing spondylitis: a randomized, double-blind trial. Arthritis Rheum 2011;63:1543–51. doi:10.1002/art.30223
      OpenUrlPubMedWeb of Science
    36. ↵
      1. Roussou E,
      2. Kennedy LG,
      3. Garrett S, et al
      . Socioeconomic status in ankylosing spondylitis: relationship between occupation and disease activity. J Rheumatol 1997;24:908–11.
      OpenUrlPubMedWeb of Science
    37. ↵
      1. Smolen JS,
      2. Braun J,
      3. Dougados M, et al
      . Treating spondyloarthritis, including ankylosing spondylitis and psoriatic arthritis, to target: recommendations of an international task force. Ann Rheum Dis 2014;73:6–16. doi:10.1136/annrheumdis-2013-203419
      OpenUrlAbstract/FREE Full Text
    38. ↵
      1. Haibel H,
      2. Rudwaleit M,
      3. Listing J, et al
      . Efficacy of adalimumab in the treatment of axial spondylarthritis without radiographically defined sacroiliitis: results of a twelve-week randomized, double-blind, placebo-controlled trial followed by an open-label extension up to week fifty-two. Arthritis Rheum 2008;58:1981–91. doi:10.1002/art.23606
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      1. Sieper J,
      2. van der Heijde D,
      3. Dougados M, et al
      . Efficacy and safety of adalimumab in patients with non-radiographic axial spondyloarthritis: results of a randomised placebo-controlled trial (ABILITY-1). Ann Rheum Dis 2013;72:815–22. doi:10.1136/annrheumdis-2012-201766
      OpenUrlAbstract/FREE Full Text
    40. ↵
      1. Ramiro S,
      2. Stolwijk C,
      3. van Tubergen A, et al
      . Evolution of radiographic damage in ankylosing spondylitis: a 12 year prospective follow-up of the OASIS study. Ann Rheum Dis 2015;74:52–9. doi:10.1136/annrheumdis-2013-204055
      OpenUrlAbstract/FREE Full Text
    41. ↵
      1. Dougados M,
      2. Demattei C,
      3. van den Berg R, et al
      . Rate and predisposing factors for sacroiliac joint radiographic progression after a two-year follow-up period in recent-onset spondyloarthritis. Arthritis Rheumatol 2016;68:1904–13. doi:10.1002/art.39666
      OpenUrl

    Footnotes

    • Handling editor Tore K Kvien

    • Contributors The authors take responsibility for the integrity of the work as a whole, from inception to published article and they should indicate that they had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    • Funding AM received a master's degree grant from the French Society of Rheumatology to perform these analyses.

    • Competing interests None declared.

    • Ethics approval The DESIR cohort was approved by an Ethics Committee (Comité de Protection des Personnes Ile de France) and all patients gave their informed consent at the inclusion on the cohort.

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