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

Disclosures

Rheumatology. 2021;60(7):3189-3198. 

In This Article

Results

Among a total of 2687 participants, 2043 were included for analysis; exclusions were due to missing questionnaires (n = 364), missing comorbidity data (n = 6) and questionnaire outside the eligible window (n = 274). The analysis population was predominantly male (67%) with mean age of 49.1 (S.D. 14.7) years. Classification criteria for AS were fulfilled by 1316 (64%). HLA-B27 status was available for 74% of participants and was positive in 79% of these cases. The mean BMI was 27.7 kg/m2. Current smoking was reported by 19% of participants, past smoking by 37% and 44% never smoked. Thirty-one per cent were in the biologic group (but had not yet commenced on treatment).

The prevalence of each comorbidity is shown in Supplementary Figure S1, available at Rheumatology online. Forty-four per cent of participants had at least one of the 14 comorbidities; 27% had one comorbidity, 10% had two and 5% had three or more (Supplementary Figure S2, available at Rheumatology online).

Table 1 compares participants with and without comorbidities. Those with comorbidities were older, had higher BMI and trend for lower educational attainment. Although there were more ever-smokers in the group with comorbidities (63 vs 50%), a larger proportion had quit (43 vs 32%). NSAID use in the preceding 6 months was less common among those with comorbidities. Participants with comorbidity also had higher disease activity and other measures of disease severity, but similar levels of inflammatory markers.

Comorbidities and Disease Activity

With comorbidity count as a continuous variable, each additional comorbidity was associated with higher BASDAI by 0.40 (95% CI: 0.27, 0.52) units and spinal pain by 0.53 (95% CI: 0.37, 0.68) units. Figure 1 shows these relationships with comorbidity count as a categorical variable (full model coefficients shown in Supplementary Table S1, available at Rheumatology online). For each additional comorbidity, ASDAS was higher by 0.09 (95% CI: 0.03, 0.15) units. Those with one or two comorbidities did not have higher ASDAS than those with none in terms of statistical or clinical significance. Comorbidity count was not associated with log-transformed CRP [β = −0.03 (back-transformed effect size −0.03 mg/dl); 95% CI: –0.07, 0.02] or log-transformed ESR [β = −0.03 (i.e. 0.97 mm/h); 95% CI: –0.12, 0.06].

Figure 1.

Association between comorbidity count and disease activity
Results shown as adjusted model coefficients with 95% CIs using participants with no comorbidities as the reference group; covariates were age, gender, BMI, smoking status, socioeconomic status and education. For example, participants with ≥3 comorbidities had 1.5 units higher BASDAI and 0.38 units higher ASDAS than those without comorbidities. ASDAS: AS disease activity score.

Independent associations between each comorbid condition and disease activity are shown in Figure 2 and Supplementary Table S2, available at Rheumatology online. Participants with depression, heart failure and peptic ulcer diseases had consistently higher disease activity than those without each of these conditions. For example, participants with depression had 0.9 units higher BASDAI and spinal pain than those without, accounting for covariates and all other comorbidities. Effect sizes were smaller for ASDAS. The only comorbidities associated with CRP and ESR were COPD and asthma, respectively; the back-transformed effect sizes for CRP (0.5 mg/dl) and ESR (0.7 mm/h) were not clinically meaningful.

Figure 2.

Association between each comorbid condition and disease activity
Results shown as adjusted model coefficients with 95% CIs compared with participants without each condition; covariates were age, gender, BMI, smoking status, socioeconomic status and education. For example, participants with heart failure (HF) had 1.7 units higher BASDAI and 0.59 units higher ASDAS than those without HF. ASDAS: AS disease activity score; COPD: chronic obstructive pulmonary disease; DM: diabetes mellitus; HF: heart failure; HTN: hypertension; IHD: ischaemic heart disease; PUD: peptic ulcer disease; TB: tuberculosis.

Comorbidities and Other Measures of Disease Severity

Comorbidity count (as continuous variable) was significantly associated with worse fatigue (β = 1.05; 95% CI: 0.76, 1.33), quality of life (β = 1.18; 95% CI: 0.90, 1.46) and functional impairment (β = 0.55; 95% CI: 0.41, 0.69). Effect size was smaller for BASMI (β = 0.22; 95% CI: 0.12, 0.33) than BASFI. Figure 3 shows these relationships with comorbidity count as a categorical variable (full model coefficients shown in Supplementary Table S3, available at Rheumatology online).

Figure 3.

Association between comorbidity count and other measures of disease severity
Results shown as adjusted model coefficients with 95% CIs using participants with no comorbidities as the reference group; covariates were age, gender, BMI, smoking status, socioeconomic status and education. For example, participants with ≥3 comorbidities had 2.0 units higher BASFI and 0.79 units higher BASMI than those without. ASQoL: AS quality of life questionnaire; BASFI: Bath AS functional index; BASMI: metrology index.

Independent associations between each comorbid condition and the four disease severity measures are shown in Figure 4 (full model coefficients shown in Supplementary Table S4, available at Rheumatology online. Participants with heart failure, depression and peptic ulcer disease had consistently worse fatigue, quality of life and functional impairment than those without, accounting for covariates and all other comorbidities. Diabetics had worse function and quality of life, while participants with stroke had higher fatigue. The only comorbidities associated with CRP and ESR were COPD and asthma, respectively; the back-transformed effect sizes for CRP (0.5 mg/dl) and ESR (0.7 mm/h) were not clinically meaningful.

Figure 4.

Association between each comorbid condition and other measures of disease severity
Results shown as adjusted model coefficients with 95% CIs compared with participants without each condition; covariates were age, gender, BMI, smoking status, socioeconomic status and education. ASDAS: AS disease activity score; ASQoL, AS quality of life questionnaire; BASFI: Bath AS functional index; BASMI: metrology index; COPD: chronic obstructive pulmonary disease; DM: diabetes mellitus; HF: heart failure; HTN: hypertension; IHD: ischaemic heart disease; PUD: peptic ulcer disease; TB: tuberculosis.

Independent Influence of Comorbidity on the Patient Global Component of ASDAS

Patient global score increased (suggesting an increase in disease severity) with the number of comorbidities, but not independently of other ASDAS components (Figure 5; full model coefficients shown in Supplementary Tables S3 and S4, available at Rheumatology online). Depression, peptic ulcer and renal diseases were significantly associated with patient global, but not when additionally adjusting for other ASDAS components.

Figure 5.

Association between the patient global score and comorbidity
Results shown as model coefficients with 95% CIs; covariates were age, gender, BMI, smoking status, socioeconomic status and education. ASDAS, AS disease activity score; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; HF, heart failure; HTN, hypertension; IHD, ischaemic heart disease; PG: patient global score; PUD, peptic ulcer disease; TB: tuberculosis.

Results from both sensitivity analyses (see Supplementary Table S5 and Supplementary Figs S3 and S4, available at Rheumatology online) did not materially change effect estimates, but precision was reduced (i.e. CIs widened) with smaller sample sizes.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....