Article Text

Download PDFPDF

Extended report
Inequity in biological DMARD prescription for spondyloarthritis across the globe: results from the ASAS-COMOSPA study
  1. Elena Nikiphorou1,2,
  2. Désirée van der Heijde1,
  3. Sam Norton2,
  4. Robert BM Landewé3,
  5. Anna Molto4,
  6. Maxime Dougados4,
  7. Filip E Van den Bosch5,6,
  8. Sofia Ramiro1
    1. 1 Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands
    2. 2 Academic Rheumatology Department, King’s College London, London, UK
    3. 3 Department of Clinical Immunology and Rheumatology, Amsterdam Rheumatology Center, Amsterdam, The Netherlands and Zuyderland Medical Center, Heerlen, The Netherlands
    4. 4 Department of Rheumatology, Paris Descartes University, Department of Rheumatology - Hôpital Cochin, Assistance Publique - Hôpitaux de Paris, INSERM (U1153): Clinical epidemiology and biostatistics, PRES Sorbonne Paris-Cité, Paris, France
    5. 5 VIB Inflammation Research Center, Ghent University, Ghent, Belgium
    6. 6 Department of Rheumatology, Ghent University Hospital, Ghent, Belgium
    1. Correspondence to Dr Elena Nikiphorou, Academic Rheumatology Department, King’s College London, Weston Education, Denmark Hill, SE1 1UL, UK; enikiphorou{at}


    Objectives The value of biological disease-modifying antirheumatic drugs (bDMARDs) in spondyloarthritis (SpA) is well recognised, but global access to these treatments can be limited due to high costs and other factors. This study explores country variation in the use of bDMARDs in SpA in relation to country-level socioeconomic factors.

    Methods Patients fulfilling the Assessment in SpondyloArthritis International Society (ASAS) SpA criteria in the multinational, cross-sectional ASAS Comorbidities in Spondyloarthritis study were studied. Current use of bDMARDs or conventional synthetic DMARDs (csDMARDs) was investigated in separate models, with multilevel logistic regression analysis, taking the country level into account. Contribution of socioeconomic factors, including country health expenditures, gross domestic product and human development index as independent country-level factors, was explored individually, in models adjusted for sociodemographic as well as clinical variables.

    Results In total, 3370 patients from 22 countries were included (mean (SD) age 43 (14) years; 66% male; 88% axial disease). Across countries, 1275 (38%) patients were bDMARD users. Crude mean bDMARD use varied between 5% (China) to 74% (Belgium). After adjustment for relevant sociodemographic and clinical variables, important variation in bDMARD use across countries remained (P<0.001). Country-level socioeconomic factors, specifically higher health expenditures, were related to higher bDMARD uptake, though not meeting statistical significance (OR 1.96; 95% CI 0.94 to 4.10). csDMARD uptake was significantly lower in countries with higher health expenditures (OR 0.32; 95% CI 0.15 to 0.65). Similar trends were seen with the other socioeconomic variables.

    Conclusions There remains important residual variation across countries in bDMARD uptake of patients with SpA followed in specialised SpA centres. This is independent of well-known factors for bDMARD use such as clinical and country-level socioeconomic factors.

    • spondyloarthritis
    • dmards (biologic)
    • dmards (synthetic)

    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 role of biological disease-modifying antirheumatic drugs (bDMARDs) in spondyloarthritis (SpA) has been extensively studied, and robust scientific evidence supports their efficacy in reducing disease activity and improving functional ability, spinal mobility and quality of life.1 bDMARDs are therefore recommended for use in the presence of active disease and following failure of two non-steroidal anti-inflammatory drugs (NSAIDs).2 However, an important barrier to their use is their high cost, which also influences the development of national guidelines and prescribing patterns.

    The use of conventional synthetic DMARDs (csDMARDs) in SpA, unlike rheumatoid arthritis (RA) and other inflammatory arthritides with peripheral joint involvement, is less well established. Currently, there is a general lack of evidence on their role in axial spondyloarthritis (axSpA),1 and the existing evidence consistently shows no efficacy,3–5 making their role debatable6 and resulting in the Assessment in SpondyloArthritis international Society (ASAS) and the European League Against Rheumatism not supporting their use in patients with only axial disease.2

    Existing literature supports inequity in bDMARD prescription in RA, both at an individual and country level,7–12 but the evidence for this is lacking in SpA. Increasing insight into patterns of treatment use across countries and potential differential access to biological drugs can help highlight potential sources of inequity and drive change through informing service delivery, refining drug reimbursement criteria and access to these treatments nationally, in line with international recommendations. This is particularly important since access and use of healthcare services that prevent and treat disease is one of the key determinants of health.13

    This study aimed to explore individual-level and country-level variation in the uptake of DMARDs in patients with SpA and unravel gaps in literature regarding how they are used and possible factors that could influence this. The ASAS COMOrbidities in SPondyloArthritis (COMOSPA) study, an international study including patients from 22 countries and initially designed to estimate the prevalence of comorbidities in SpA,14 provided an ideal setting to answer these questions.


    Study design and patient recruitment

    ASAS-COMOSPA is a multicentre cross-sectional observational study with 22 participating countries across four continents (Africa, America, Asia and Europe).14 Consecutive patients (age 18 years or over) with a clinical diagnosis of SpA according to the treating rheumatologist, either axial or peripheral, were included in ASAS-COMOSPA, provided they were able to understand and complete the questionnaires. For the present study, analyses were restricted to patients fulfilling the ASAS criteria for SpA, either axial or peripheral.15 Written informed consent was obtained from all subjects before enrolment.

    Data collection

    Data collection in the ASAS-COMOSPA ranged from patient demographic variables to disease-related variables and treatment data, including: treatment with NSAIDs with computation of the ASAS NSAID score (0–400)16 reflecting NSAID use over the past 3 months and current and past use of csDMARDs and bDMARDs (see below).

    Outcome measures

    The main outcome of interest was current bDMARD uptake, studied as a binary variable to indicate current bDMARD use versus all other (including csDMARD use and/or NSAIDs). In addition, current csDMARD uptake as a binary variable to indicate current csDMARD use versus all other (including bDMARD use with or without csDMARDs and/or NSAID use) was also examined in separate models as another outcome measure.

    Individual-level variables

    Variables of interest potentially influencing the uptake of DMARDs, aside from age and gender, included sociodemographic factors such as educational status (secondary and university education vs primary education); HLA B27 status (positive vs negative); measures of disease activity such as the Ankylosing Spondylitis Disease Activity Score calculated with C reactive protein; measures of functional ability (Bath Ankylosing Spondylitis Functional Index (range 0–10); presence of axial versus peripheral disease (yes for axial disease); radiographic sacroiliitis (yes vs no); presence of peripheral enthesitis, dactylitis or extra-articular manifestations (uveitis, psoriasis or inflammatory bowel disease) (yes vs no); and comorbidity burden using the Rheumatic Disease Comorbidity Index (range 0–9).17

    Country-level variables

    Country socioeconomic variables were studied as the main independent variables of interest and included: country health expenditures per capita18 (adjusted for purchasing power parity (PPP), measured in international dollars); gross domestic product (GDP)19 (adjusted for PPP, measured in international dollars); Gini index,20 21 as a measure of income inequality across a country (range 0 (absolute equality)−100 (absolute inequality)); and human development index (HDI),22 a composite measure of average achievement in key dimensions of human development used to rank countries based on their performance in these. These variables were split into tertiles with the top two compared with the bottom tertile in regression analyses: for country health expenditures, GDP and Gini and high/medium versus low. For HDI, an external classification system was used22 as opposed to creating a new dichotomisation, with categories compared being high/very high versus medium. All country-level socioeconomic variables are presented in online supplementary table 1. The country health expenditures variable was a priori chosen as the main independent variable of interest, as the outcome refers to the uptake of a drug, falling into health expenditures. Therefore, we hypothesised that country health expenditures would be the most relevant socioeconomic variable in the context of health spending and a good reflection of country wealth.

    Supplementary file 1

    Data analysis

    Multilevel modelling analyses were conducted in order to account for patients being recruited from different countries. Multilevel models take into account the dependency of the observations, in this instance by accounting for the two-level structure in the data, namely patients at the ‘lower’ level are nested within countries at the ‘higher’ level.23 Multilevel mixed effects logistic regression models with random intercept for country were constructed with current use of bDMARDs and current use of csDMARDs as the dependent variables, in separate models. ORs and 95% CIs were estimated. Variations in impact of patient-level sociodemographic variables (age, gender and educational status) on DMARD use across countries were first tested by incorporating random slopes for the variable, which is comparable with testing for interactions in a simple regression model. The effect of level of education was found to vary significantly (P<0.001) across countries in relation to bDMARD uptake; therefore, education was included with a random slope in multivariable models where bDMARD was the outcome to control for potential confounding at the country as well as individual level. Potential confounders were entered in the models in a manual forward procedure (cut-off P<0.05) provided they were meaningful in the univariable analyses (defined as P<0.10) or if considered clinically relevant. In a final step, the contribution of country health expenditures, GDP, Gini and HDI as independent country-level factors was individually explored in models adjusted for sociodemographic (age, gender and education level) as well as clinical variables (presence of axial vs peripheral disease, disease activity, sacroiliitis on X-ray, history of extra-articular manifestations, total NSAID score and past csDMARD/bDMARD use) known to determine bDMARD use (or csDMARD use, respectively) in SpA. All analyses were conducted with the statistical software Stata V.13.


    Patient, disease characteristics and treatment

    From a total of 3984 patients included in ASAS-COMOSPA across 22 countries, 3370 (85%) fulfilled the ASAS SpA criteria for axial or peripheral disease and were included in this study. The majority of patients were male (66%); mean age was 43 years (SD 14), mean disease duration 8.4 years (SD 9.5) and 88% had axial disease. Table 1 summarises the patient demographics, clinical characteristics and type of treatment used. Results by individual country are shown in online supplementary table 2. Across countries, 1275 (38%) patients were bDMARD users, 1168 (35%) were csDMARD users (25% without bDMARDs). Crude mean bDMARD and csDMARD uptake varied considerably across countries (see figure 1).

    Supplementary file 2

    Figure 1

    bDMARD uptake (%) by country. Crude and adjusted percentage use shown along with 95% CI based on models with socioeconomic, sociodemographic and clinical variables. Countries ranked based on health expenditure: low (left) to high (right). bDMARD, biological disease-modifying antirheumatic drugs.

    Table 1

    Patient demographics, clinical characteristics and treatment in patients with SpA fulfilling the ASAS classification criteria

    bDMARD uptake

    Table 2 shows the model with bDMARD uptake as the outcome. Higher country health expenditure was associated with higher bDMARD uptake (OR 1.96; 95% CI 0.94 to 4.10), though without reaching statistical significance. In the same models, past bDMARD/csDMARD use was associated with almost double odds of using bDMARDs. Similarly, male gender, presence of axial (vs peripheral) disease, sacroiliitis on X-ray and presence of extra-articular manifestations were all significantly associated with higher bDMARD use. The results also suggest an association between lower disease activity with lower bDMARD use, likely to be a reflection of the cross-sectional nature of the study (ie, simply an observation of less disease activity in those already on bDMARDs). Figure 1 shows the crude and adjusted percentage of bDMARD uptake by country. The model demonstrated significant variation in bDMARD use by country (P<0.001) despite full adjustment.

    Table 2

    Uptake of bDMARDs: association with sociodemographic, clinical and treatment variables as well as indicators of the country socioeconomic welfare

    csDMARD uptake

    Table 3 shows the model with csDMARD uptake as the outcome. Higher country health expenditure was associated with lower csDMARD uptake (OR 0.32; 95% CI 0.15 to 0.65). The results of the csDMARD model are complimentary to those of the bDMARD model, with the same variables demonstrating an association with csDMARD uptake in the opposite direction to those of bDMARD uptake. In other words, male gender, axial disease and sacroiliitis on X-ray and past csDMARD use were all significantly associated with lower csDMARD use. Higher disease activity was associated with higher csDMARD use, again likely to be a reflection of the cross-sectional nature of the study (ie, higher disease activity in those using csDMARDs). Figure 2 shows the crude and adjusted percentage of csDMARD uptake by country. A significant variation across countries was also seen in relation to csDMARD uptake (P<0.001) and also independent of adjustment for sociodemographic, clinical and socioeconomic relevant variables.

    Figure 2

    csDMARD uptake (%) by country. Adjusted and crude percentage use shown along with 95% CI based on models with socioeconomic, sociodemographic and clinical variables. Countries ranked based on health expenditure: low (left) to high (right). csDMARD, conventional synthetic disease-modifying antirheumatic drugs.

    Table 3

    Uptake of csDMARDs: association with sociodemographic, clinical and treatment variables as well as indicators of the country socioeconomic welfare

    Other country-level socioeconomic variables

    Across other socioeconomic variables studied, the only significant association in univariable analyses was between HDI and csDMARD uptake. Replacing country health expenditures in the final adjusted models with other country-level socioeconomic variables revealed higher use of bDMARDs and lower use of csDMARDs with higher GDP and HDI, although significance was only reached for GDP and csDMARD use (OR 0.44; 95% CI 0.21 to 0.91) (table 4). Higher country-income inequality as measured by Gini was associated with lower bDMARD than csDMARD uptake, although no statistical significance was reached (table 4).

    Table 4

    Relationship between country-level socioeconomic factors and bDMARD and csDMARD use, all tested individually in separate models (each cell represents a different model)


    The ASAS-COMOSPA study enabled the systematic study of bDMARD and csDMARD uptake across 22 countries. It demonstrates important residual variation, which is not explained by sociodemographic and clinical characteristics. The study suggests that country-level socioeconomic indicators may in part, but not entirely, explain some of the differences. The csDMARD findings are supportive of the bDMARD results, highlighting that higher country welfare seems to be associated with higher bDMARD use (although not reaching statistical significance), independent of all other characteristics including country of residence, and with lower csDMARD use. Given the lack of evidence for efficacy of csDMARDs in axSpA1 and the available evidence consistently showing no efficacy,2–6 this reflects an unjust selection of treatment for patients in countries of lower socioeconomic welfare, based on decisions other than clinical indication. bDMARD use was almost double in countries with higher compared with lower country health expenditures. Although not reaching statistical significance, the effect is of interest, since power to detect country-level predictors is driven largely by the number of countries. The number of countries included in ASAS-COMOSPA, though impressive for a multinational study with the logistic challenges it represents, is relatively small in statistical terms and a limiting factor when analysing country-level variables. This, in turn, is reflected in a lack of power to identify potentially significant relationships.

    To date, only few studies have systematically studied access to biologicals across countries and these have been mainly in RA.7–12 Our study observations find support in the existing literature of bDMARD use in RA that suggests country-level socioeconomic factors to play a role.10 12 24 In particular, existing evidence shows that patients living in countries with a higher welfare have lower disease activity states, likely to be at least in part mediated by a higher likelihood of receiving bDMARDs.12 The high costs of these drugs have undoubtedly influenced reimbursement and also national recommendations and guidelines across countries, in order to regulate access to these treatments while keeping a balance between clinical and economic demands.25 26 Indeed, costs of bDMARDs vary widely by country, driven by socioeconomic welfare among other factors9 with countries of lower socioeconomic welfare demonstrating stricter eligibility criteria for bDMARDs.11

    The existence of international recommendations in SpA2 encourage comparable management in these patients. In fact, evidence suggests that most national recommendations follow the international ASAS recommendations and despite some countries requiring, for example, additional objective signs of inflammation and/or more pretreatment, limiting access, general consensus exists about the use of, for example, tumour necrosis factor-inhibitor therapies.27 Still, there could be ‘hidden’ barriers across individual countries limiting access to these drugs, ranging from differences in the funding of healthcare provision, to local/regional variation in budget availability and feasibility of access to these more expensive, although more effective treatments, through to differences in guideline interpretation and personal approach as well as preference by the treating rheumatologist. It may be, for example, that knowledge about the potential side effects of bDMARDs poses resistance to their use by some individuals, who may in turn seek out to alternative treatments. This may explain the differences observed even between countries with comparable health expenditures. We can only speculate on the reasons for the residual degree of variation in bDMARD uptake in our study, despite adjustment for patient, disease and country-level characteristics. It is also possible that patient selection at inclusion into the study may have played a role in these observations. For example, preferential review of patients on bDMARDs by some centres would not provide an accurate reflection of the wider practice at a specific clinical setting and less so across the entire country. Furthermore, it is possible that not always consecutive patients may have been selected for inclusion into the study. The fundamental issue though remains that, assuming the patient needs for bDMARD use are similar across countries, differential access to these treatments raises concerns regarding the risk of inequity.

    Male patients, presence of axial disease, sacroiliitis on X-ray and presence of extra-articular disease were all associated with higher bDMARD use. In the csDMARD model, these associations were reversed and therefore supportive of the bDMARD findings. These observations are reassuring, since all these factors are indicators of worse disease or better response and justify higher bDMARD use.2 28 29

    The study has some important limitations. First, selection bias cannot be excluded and the uptake of bDMARDs in the group of patients included per country may not be fully representative of the general bDMARD uptake across all SpA patients. More specifically, the study has been conducted in centres that are associated with ASAS, and this may be a bias towards higher bDMARD prescription, independent of the country and related socioeconomic factors. Better selection of patients for bDMARD use is possible in ASAS centres. This reflects potential sources of bias to the findings of the study. However, consecutive patients were included in the study, and the disease characteristics of the population studied is reflective of a typical SpA population. Second, it was not possible to explore all possible reasons for barriers to access of bDMARDs, and as mentioned above, explanations for the residual variation seen in bDMARD use after adjusting for socioeconomic, sociodemographic and clinical variables remain speculative. The aim, however, was to investigate whether differential access could be a problem and potentially lead to inequities. Further research should unveil possible other explanations for treatment choices. Furthermore, the cross-sectional nature of ASAS-COMOSPA precludes the study of causal links; instead, it only allows for associations to be seen. Finally, the cross-sectional nature of the analysis prevents the adjustment of disease activity before the start of bDMARDs, which is another important limitation.

    Important strengths of the study include the large patient numbers and the uniqueness of ASAS-COMOSPA as one of the largest multinational SpA datasets to date, which includes a wealth of information ranging from sociodemographic to disease-related clinical and radiographic measures of disease as well as country-level macroeconomic data. The study population is typical and representative for SpA, characterised by predominantly male patients with an average age in the early 40s. The occurrence of disease at the peak of the productive lifespan of young individuals30 31 with the known considerable impact on work ability32 makes it imperative that access to treatments that are known to be effective in suppressing inflammation is feasible and unrestricted. This, alone, makes our study particularly relevant.

    In conclusion, this study provides insights into complex contributions between patient and disease-related factors and country-level socioeconomic factors, raising concerns regarding equity in access to effective (biological) treatments in SpA. The findings suggest unequal and unjust selection of treatment for SpA independent of clinical indication, an observation that necessitates urgent attention on the health equality and public health agenda.


    The COMOSPA study was conducted under the umbrella of the International Society for Spondyloarthritis Assessment (ASAS). We would like to thank all ASAS-COMPOSPA collaborators.



    • Handling editor Josef S Smolen

    • Twitter @ElenaNikiUK

    • Contributors EN wrote a first draft of the manuscript, which was then reviewed and edited by SR, RBML, DvdH and SN. AM, MD, FEvdB and SN provided advice and support on the statistical analyses, and all coauthors contributed to further revisions of the manuscript and confirmed approval prior to final submission.

    • Funding The COMOSPA study was conducted with the financial support of Abbvie, Pfizer and UCB, who provided an unrestricted grant to ASAS to fund the study.

    • Disclaimer The funders did not have any role in the design or conduct of the study. This ancillary study did not receive any funding, and the sponsors of COMOSPA did not have any interference with this current study.

    • Ethics approval The study was conducted according to guidelines for good clinical practice in all countries with all local ethics committees approving the ASAS-COMOSPA study protocol.

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

    • Collaborators Fadoua Allali MD, Morocco; Raquel Almodovar Gonzàlez MD, Spain; Elena Alonso Blanco-Morales MD, Spain; Alejandro Alvarellos MD, Argentina; Maria Aparicio Espinar MD, Spain; Pamir Atagunduz MD, Turkey; Pauline Bakker MD, Netherlands; Juan C. Barreira MD. Argentina; Leila Benbrahim MD, Morocco; Bahia Benchekroun MD, Morocco; Alberto Berman MD, Argentina; Juergen Braun MD, Germany; Alain Cantagrel MD PhD, France; Roberto Caporali MD, Italy; Pedro Carvalho MD, Portugal; Gustavo Casado MD, Argentina; James Cheng-Chung Wei MD, PhD, Tawian; Francisco Colombres MD, Argentina; Eugenio del Miguel Mendieta MD PhD, Spain; Juan D. Diaz-Garcia MD, Mexico; Michel De Bandt MD PhD, France; Vanesa Duarte MD, Argentina; Cristina Fernandez Carballido MD, Spain; Mari Cruz Fernandez Espartero MD, Spain; Manuel Fernandez-Prada MD, Spain; Rene-Marc Flipo MD PhD, France; Pilar Font Ugalde MD. PhD, Spain; Philippe Gaudin MD PhD, France; Philippe Goupille MD, France; Dolors Grados Cànovas MD, Spain; Jordi Gratacós Masmitjà MD PhD, Spain; Vittorio Grosso MD, ITALY; Naomi Ichikawa, MD Japan; Hisashi Inoue MD, Japan; Yuko Kaneko MD PhD, Japan; Taku Kawasaki MD PhD, Japan; Shigeto Kobayashi MD, Japan; Manjari Lahiri MD, Singapore; Hernàn Maldonado-Ficco MD, Argentina; Marhadour MD, France; Alejandro Martínez MD, Argentina; Kazuo Matsui MD, Japan; Ramón Mazzuchelli Esteban MD, Spain; Corinne Micelli MD PhD, France; Chisun Min MD, Japan; Mitsuhiro Morita MD PhD, Japan; Juan Mulero Mendoza MD PhD, Spain; Jose Raul Noguera Pons MD, Spain; Masato Okada MD, Japan; Alberto Ortiz MD, Argentina; Jon Packham DM FRCP, UK; Gisela Pendón MD, Argentina; Dora Pereira MD, Argentina; José A Pereira da Silva MD, Portugal; Fernando Pimentel-Santos MD, Portugal; Hanan Rkain, MD MOROCCO; Oscar Rillo MD, Argentina; Carlos Rodriguez Lozano MD, Spain; Adeline Ruyssen-Witrand MD PhD, France; Adriàn Salas MD, Argentina; Carlos Salinas-Ramos MD, Mexico; Amelia Santosa MD, Singapore; Alain Saraux MD PhD, France; Raj Sengupta FRCP PGCME, UK; Stefan Siebert PhD, UK; Martin Soubrier MD PhD CHU, France; Caroline Spiegel, Germany; Carmen Stolwijk MD, Netherlands; Kurisu Tada MD, Japan; Naoho Takizawa MD, Japan; Yoshinori Taniguchi MD PhD, Japan; Atsuo Taniguchi MD PhD, Japan; Chung Tei Chou MD, Taiwan; Lay-Keng Teoh Singapore; Tetsuya Tomita MD PhD, Japan; Wen-Chan Tsai MD, PhD, Taiwan; Shigeyoshi Tsuji MD PhD, Japan; Olga Tsyplenkova, Germany; Astrid van Tubergen MD PhD, Netherlands; Kiana Vakil-Gilani BS, MPH, USA; Rafael Valle-Oňate MD, Colombia; Gaelle Varkas MD, Belgium; Virginia Villaverde MD, Spain; Ai Yap Singapore; Pedro Zarco Montejo MD PhD, Spain.