Association Between Comorbidities and Disease Activity in Axial Spondyloarthritis

Results From the BSRBR-AS

Sizheng Steven Zhao; Gareth T. Jones; Gary J. Macfarlane; David M. Hughes; Robert J. Moots; Nicola J. Goodson


Rheumatology. 2021;60(7):3189-3198. 

In This Article


The British Society for Rheumatology Biologics Register for AS (BSRBR-AS) is a UK-wide prospective cohort study of biologics-naïve patients fulfilling the ASAS criteria for axial SpA. Patients were recruited between December 2012 and December 2017 into two groups: a 'biologic' group [those starting biologic DMARDs (bDMARDs)] and a 'non-biologic' group (those not). The study protocol[14] and cohort characteristics have been previously published.[15] This analysis focussed on baseline (cross-sectional) data before the biologic group started bDMARDs. This analysis used the study dataset of December 2018.

Participating centres obtained physician diagnosed comorbidity data from medical records. The list of comorbidities included: myocardial infarction, angina, heart failure, stroke, hypertension, diabetes, asthma, chronic obstructive pulmonary disease (COPD), peptic ulcer disease, liver disease, renal disease, depression, cancer, tuberculosis (TB) and demyelinating disease. These conditions were selected through a consensus meeting of clinicians and researchers, based on commonly recorded comorbidities in routine practice. Comorbidity status was defined at baseline. Myocardial infarction and angina were combined as ischaemic heart disease (IHD) for this analysis. Extra-articular manifestations of axSpA (uveitis, psoriasis and IBD) were considered disease features rather than comorbidities, given that they share pathogenesis with axSpA and form part of the classification criteria (thereby determining study inclusion).[16,17]

Questionnaires collected PROMs, highest educational attainment and smoking status. Baseline visits and questionnaires did not necessarily coincide; we included questionnaires within 1 year before or after the baseline visit for the non-biologic group, and 1 year before to 7 days after for the biologic group (the 'eligible window'). Disease activity was assessed using the Bath AS Disease Activity Index (BASDAI), AS Disease Activity Score (ASDAS) and spinal pain numerical rating scale; inflammation using CRP (mg/dl) and ESR (mm/h); functional impairment using Bath AS Functional (BASFI) and Metrology Indices (BASMI); fatigue using the Chalder Fatigue Scale Likert scale (CFQ), which has a range of 0–33 with higher scores indicating greater levels of fatigue;[18] and quality of life using the AS quality of life questionnaire (ASQoL), which has a range of 0–18 with higher scores indicating poorer quality of life. All patient-reported indices were collected at the same time. They are collectively referred to as measures of disease severity throughout the text. To provide context for interpretation of results, minimal clinically important difference in BASDAI is around 1 unit, pain numerical rating scale 1.6 units, ASDAS 1.1 units, BASFI 0.6 units, ASQoL (meaningful deterioration) 1 unit, CFQ 2.3–3.3 units, and undefined for the remaining indices.[19–23]

ASDAS was calculated using the formula 0.12 × Back Pain + 0.06 × Duration of Morning Stiffness + 0.07 × Peripheral Pain/Swelling + 0.11 × Patient Global + 0.58 × ln(CRP + 1); that is, questions 2, 3 and 6 from BASDAI, plus patient global which asks 'How active was your spondylitis on average during the last week?'.[8,9] Where CRP was not available, ASDAS was calculated using ESR.[8,9]

Covariates were determined a priori based on discussion and causal diagrams,[24] including age, gender (female as referent), BMI, smoking status (ever/never), socioeconomic status (as continuous variable) and educational attainment (as dummy variables). Socioeconomic status was approximated using post-code derived Index of Multiple Deprivation that related to the specific country of residence within the UK, quintile 1 representing the top 20% most deprived areas and quintile 5 the least deprived.[17,18] Smoking was categorized as ever and never, since comorbidities will influence smoking cessation behaviour. Similarly, use of NSAIDs in the past 6 months is an intermediate variable, and thus excluded as a covariate (see Supplementary Data S1, available at Rheumatology online, for causal diagrams and justification).

Ethical approval was obtained from the National Research Ethics Service Committee North East—County Durham and Tees Valley (reference 11/NE/0374) and informed consent was obtained from all participants.


Descriptive statistics were used to compare participants with and without comorbidities (Student's t-test and the Wilcoxon rank-sum test for continuous variables, Fisher's exact test for categorical variables). The count of 14 comorbidities was entered as a continuous variable into linear models to describe its association with each disease severity measure, adjusting for the above covariates. To test for non-linear relationships and to facilitate interpretation, we also categorized comorbidity count as 0, 1, 2 or ≥3 [only 32 patients (1.6%) had four or more comorbidities]. CRP and ESR were transformed using ln(CRP + 1) and ln(ESR).

We also examined the independent contribution of individual comorbidities to each disease severity measure by adding all 14 comorbidities into linear models, adjusting for the same covariates. Individual conditions may be closely related to others (e.g. hypertension and IHD); we therefore examined variance inflation factors[25] to check for multicollinearity. Correction for multiple testing was not performed since dependent variables all measure the same underlying construct of disease severity.

To examine whether comorbidity count or individual comorbid conditions independently inflated the patient global score, we repeated the above analyses for patient global and comorbidity count or individual comorbidities, but additionally adjusting for other components of the ASDAS (three questions and CRP). Throughout, model coefficients and 95% CIs are displayed graphically with detailed results provided in the Supplementary Data, available at Rheumatology online. Complete case analysis was used throughout with no imputation.

Sensitivity Analyses

Some comorbidities, including heart failure, cancer, TB and demyelinating diseases, are routinely sought during the workup for TNF inhibitor therapy. It is therefore possible that these comorbidities are more prevalent in the biologic group due to differential ascertainment alone. We repeated all analyses in only the non-biologic cohort. For analyses of patient global (third aim), successful adjustment for other components of ASDAS required adequate covariate overlap between comparison groups; extrapolating beyond overlap can introduce bias. We therefore matched patients on all covariates [gender, ever-smoked, education, index of multiple deprivation, quintiles of age and BMI, and tertiles of the three ASDAS questions and ln(CRP + 1)] in sensitivity analyses (using coarsened exact matching).[26] All analyses were performed using Stata version 13 (StataCorp, College Station, TX, USA).