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


Patient-reported axSpA disease activity increased with the number of comorbidities in this cross-sectional study. Unlike BASDAI and spinal pain, ASDAS was not associated with comorbidity count or individual comorbidities at a clinically meaningful effect size. Although patient global score was influenced by coexisting morbidities, they did not inflate ASDAS through the patient global score independently of other ASDAS components. When making treatment decisions, clinicians should be mindful of the potential impact of comorbidities on patient reported measures of disease activity and other measures of disease severity. Disease activity should be assessed using ASDAS when comorbidities are present.

A key strength of this study is its large sample size from a broad range of rheumatology centres, for whom a wide range of disease measures were collected. Ascertainment of comorbidities was robust, using physician diagnoses from medical records. There were also limitations. Selection of comorbidities was not tailored for this secondary analysis; therefore some (e.g. neurological or infectious) comorbidities were not compared. However, included comorbidities were broadly representative of important diseases when compared with prior axSpA research.[12] Low prevalence of some conditions (e.g. heart failure, liver and demyelinating diseases) meant that their effect estimates had significant uncertainty. Severity (e.g. for heart failure) and more granular description (e.g. cancer type) for comorbidities were not available but would have provided useful information. Some comorbidities (TB, heart failure, cancer and demyelinating diseases) are of special interest when considering TNF inhibition therapy; they may be recorded more systematically in patients. However, restricting analyses to the non-biologic cohort did not meaningfully change the results. Lastly, wording of the patient global question can be highly variable.[27] For example, the patient global in DAS28 can be worded to assess arthritis-related disease activity or global health (the recommended phrasing in RA is 'Considering all the ways your arthritis has affected you, how do you feel your arthritis is today?'). In our data, patient global was specific to SpA activity in the past week; therefore, other versions may not be equally robust in the presence of comorbidities.

Our results complement those from the ASAS-COMOSPA study, in which comorbidity burden (assessed using the Rheumatic Disease Comorbidity Index (RDCI)) was associated with worse BASFI, quality of life (EuroQol) and work-related outcomes, despite using a slightly different list of comorbid conditions.[5] We additionally demonstrated that BASMI—a physician derived outcome often considered objective—is also significantly associated with comorbidity count, albeit with smaller effect sizes than BASFI. Importantly, we showed ASDAS to be comparatively robust to the presence of coexisting morbidities. Unlike BASDAI and spinal pain, the relationship between comorbidity count and ASDAS was not completely linear, which may be explained by the weighting of ASDAS components. ASDAS was not significantly different between participants without comorbidities and those with one or two, but the effect size was proportionately larger for ≥3 conditions. Nevertheless, a mean difference of 0.38 between those with ≥3 conditions and none is not clinically significant (meaningful change = 1.1). Our use of an unweighted comorbidity count was preferable, since RDCI was weighted for inpatient outcomes (i.e. mortality, hospitalization, disability and costs) that may not be appropriate to study axSpA-specific measures. It also allowed us to examine the relationship between comorbidity and outcomes in more detail: depression, diabetes, heart failure, peptic ulcer and renal diseases were the most significant contributing conditions.

Neither the ASAS-COMOSPA nor our cross-sectional study could establish causal relationships. For example, heart failure is likely to cause functional impairment, but participants with heart failure may also have reduced access to treatment (NSAIDs and TNFi) to reduce functional decline. Similarly, renal and peptic ulcer diseases can be contraindications for symptomatic control with NSAIDs, but may also result from long-term need for NSAIDs due to active disease. These are limitations arising from the cross-sectional design and lack of historical NSAID data. Results from our analysis do, however, suggest that outcomes such as ASDAS and BASMI are less influenced by comorbidities than patient-reported measures.

In RA, patient global increased with the number of comorbidities, independently of tender/swollen joint count, CRP and physician global.[11] Among axSpA participants in this study, patient global was significantly associated with comorbidity count, but not independently so when adjusting for other ASDAS components. This adds further reassurance and support for the use of ASDAS, particularly when comorbidities are present and likely to influence other outcome measures. Further longitudinal studies are needed to assess whether comorbidities influence treatment response as measured by different outcomes. Longitudinal studies on comorbidities as 'exposure' (rather than outcome) in axSpA are scarce. Preliminary results from the BSRBR-AS suggest that comorbidity may be one of very few potentially modifiable predictors of treatment response.[28]

In summary our results suggest that patient-reported axSpA measures are influenced by coexisting morbidities. This is important for routine practice as around half of all patients have at least one comorbidity. ASDAS seems to be less vulnerable to the presence of comorbidities. ASDAS was not disproportionately inflated by the patient global score as was shown for DAS28 in rheumatoid arthritis. In routine clinical practice, clinicians should consider additionally collecting ASDAS to assess disease activity in patients with multiple comorbidities, including but not limited to depression. Further studies should examine the impact of comorbidity burden on longitudinal disease severity and response to treatment.