Abstract
Objective. To quantify the relationship between Stanford Health Assessment Questionnaire (HAQ) disability and arterial stiffness in patients with rheumatoid arthritis (RA).
Methods. A consecutive series of 114 patients with RA but without overt arterial disease, aged 40–65 years, were recruited from rheumatology clinics. A research nurse measured blood pressure (BP), arterial stiffness (heart rate-adjusted augmentation index), fasting lipids, glucose, erythrocyte sedimentation rate (ESR), and rheumatoid factor (RF). A self-completed patient questionnaire included HAQ, damaged joint count, EuroQol measure of health outcome, and Godin physical activity score. Multiple linear regression (MLR) adjusted for age, sex, smoking pack-years, cholesterol, mean arterial BP, physical activity, daily fruit and vegetable consumption, arthritis duration, ESR, and RA criteria.
Results. Mean age was 54 years (81% women) with a median HAQ of 1.13 (interquartile range 0.50; 1.75). Median RA duration was 10 years, 83% were RF-positive, and median ESR was 16 mm/h. Mean arterial stiffness was 31.5 (SD 7.7), BP 125/82 mm Hg, cholesterol 5.3 mmol/l, and 24% were current smokers. Current therapy included RA disease-modifying agents (90%), prednisolone (11%), and antihypertensive therapy (18%). Arterial stiffness was positively correlated with HAQ (r = 0.42; 95% CI 0.25 to 0.56). On MLR, a 1-point increase in HAQ disability was associated with a 2.8 increase (95% CI 1.1 to 4.4; p = 0.001) in arterial stiffness. Each additional damaged joint was associated with a 0.17 point increase (95% CI 0.04 to 0.29; p = 0.009) in arterial stiffness. The relationship between EuroQol and arterial stiffness was not statistically significant.
Conclusion. In patients with RA who are free of overt arterial disease, higher RA disability is associated with increased arterial stiffness independently of traditional cardiovascular risk factors and RA characteristics.
Rheumatoid arthritis (RA) and atherosclerosis share similar pathophysiological features of chronic inflammation1. It has been suggested that “accelerated atherosclerosis” is an extraarticular feature of RA attributable to chronic inflammation2. Patients with RA appear to be at an increased risk of cardiovascular (CV) death compared with the general population. A recent metaanalysis of 7 high-quality cohort studies reported a pooled increased relative risk of CV death in patients with RA of 21% (95% CI 6% to 39%)3. However, this increased CV risk may be attributable to patients with RA having a greater exposure to traditional CV risk factors (such as smoking, dyslipidemia, hypertension, and physical inactivity) than the general population4.
If chronic inflammation is the driving force behind both atherosclerosis and RA-related disability, then measurement of arthritis-related disability may be useful in assessment of the future risk of CV events in patients with RA. We hypothesized that patients with RA who have more severe inflammatory joint disease would have both a higher level of arthritis-related disability and a higher level of arterial dysfunction. The most widely used and validated disease-specific quality of health measure for RA is the Stanford Health Assessment Questionnaire (HAQ) disability index5,6. Three large prospective cohort studies have demonstrated that the HAQ disability score is an independent predictor of overall mortality7–9. It has yet to be demonstrated whether the HAQ is predictive of CV death in patients with RA.
Several studies have assessed arterial stiffness in patients with RA using the SphygmoCor device (AtCor Medical Ltd., West Ryde, NSW, Australia)10–18. This device permits noninvasive pulse wave analysis (PWA) using applanation tonometry at the radial artery and allows rapid clinical assessment of central aortic pressures without the need for cardiac catheterization19–22. In peripheral arteries the outgoing systolic pulse wave is reflected back toward the heart, where it augments the central aortic pressure in late systole. The augmentation pressure (AP) is the amount by which this reflected arterial pulse increases the central aortic pressure. The augmentation index (AIX) is the AP expressed as a proportion of the central aortic pulse pressure (AP/PP). Augmentation index is determined by left ventricular ejection, pulse wave velocity, and peripheral arterial resistance, and is also related to endothelial dysfunction19,20. It is considered by many researchers to be a useful composite index of arterial stiffness23. AIX is predictive of adverse cardiac events and may be a useful proxy for subclinical atherosclerosis24,25.
Patients with RA are at increased risk of CV death, and chronic inflammation may contribute directly to both arthritis-related disability and atherosclerosis. The HAQ disability score may be useful in the assessment of subclinical atherosclerosis in patients with RA, but no previous study has adequately assessed this. Consequently our primary aim was to quantify the relationship between arterial stiffness and HAQ disability in patients with RA who are free of overt arterial disease.
MATERIALS AND METHODS
We recruited adults aged 40–65 years with a rheumatologist’s diagnosis of RA from a consecutive series of patients attending rheumatology outpatient clinics in Aberdeen, Scotland. We identified and excluded patients with overt arterial disease (angina, myocardial infarction, stroke, transient ischemic attack, arterial revascularization, and peripheral arterial disease) using an initial patient questionnaire, detailed medical record review, and resting 12-lead electrocardiogram (ECG). We also excluded patients with atrial fibrillation, heart failure, and valvular heart disease. No participants had had any infection or been immunized within the previous 2 weeks.
Patients attended for a standardized clinical assessment in the morning, having fasted overnight and avoided tobacco, alcohol, and caffeine. Participants rested supine in a quiet room for at least 10 minutes, before undergoing PWA according to current guidelines26. Blood pressure (BP) was measured 3 times at the right brachial artery using a validated oscillometric device (IntelliSense BP monitor HEM-757, Omron, Kyoto, Japan)27. PWA was also done 3 times using the SphygmoCor device (SCOR v. 7.01) with a handheld tonometer probe at the right radial artery. Analysis is based on the mean of the 3 BP and PWA measurements. The SphygmoCor device employs a validated generalized transfer function to convert the peripheral arterial pulse wave, recorded noninvasively at the radial artery, into the equivalent central aortic arterial pulse wave19,20,23. It produces both an unadjusted and adjusted AIX (“standardized” to a heart rate of 75 bpm: AIX@75). Since AIX is known to vary with heart rate, AIX@75 was used for the main analysis28.
All PWA measurements were made by a single skilled operator with a high level of repeatability [Bland-Altman limits of agreement: AIX@75 within-observer 0.1 (SD 8.0); between-observer 1.0 (SD 3.9)]29. All PWA recordings were required to have a quality index (QI%, based on average pulse height, pulse height variation, and diastolic variation) of at least 95%. Following PWA, a fasting venous blood sample was obtained for lipid profile, glucose, erythrocyte sedimentation rate (ESR; Westergren method), and rheumatoid factor (RF). Standing body weight and height were measured (with shoes, socks, and bulky clothing removed) using a single set of electronic digital scales with combined stadiometer (Seca, model Delta 707, Hamburg, Germany).
Health status was measured using a self-reported patient questionnaire, completed on the same day as PWA. This included both a standardized disease-specific instrument (HAQ) and a generic health instrument (EuroQol). The HAQ asks about health over the previous week in relation to dressing/grooming, arising, eating, walking, hygiene, reach, grip, and activities; additional questions concern the use of aids and the need for help from others5. EuroQol asks about health on the day of assessment in 5 areas: mobility, self-care, usual activities, pain-discomfort, and anxiety-depression30. The EuroQol, a brief generic quality of health instrument, has also been widely used and validated in patients with RA30,31. Patients completed a damaged joint count (joints replaced or permanently damaged with limited movement or deformity as a result of arthritis on 45-joint mannequin); and rated their overall general health and overall arthritis-related health on a 100-point linear scale (visual analog scale, VAS) from best health (0) to worst (100). Participants self-completed lifestyle questions concerning smoking, diet, alcohol, and exercise (Godin physical activity score)32.
Our principal interest was in the relationship between disease-related disability and arterial stiffness. The English (UK) versions of the HAQ and EuroQol were scored using standard methods without any imputation required for missing data5,6,30. Since AIX varies by heart rate, it was standardized to a rate of 75 bpm28. The EuroQol utility score was multiplied by 100 to make it more easily comparable to the other two 100-point scales (overall health and overall arthritis).
The number of patients recruited permitted 11 independent variables (√114) to be included in multivariable analysis. Multiple linear regression (MLR) was used to simultaneously adjust for 10 independent variables: patient characteristics (age, sex); major CV risk factors (smoking pack-years, fasting cholesterol, mean arterial BP); features of arthritis [study ESR, duration of arthritis, American College of Rheumatology (ACR) RA criteria33]; and dietary and lifestyle factors (daily fruit and vegetables, Godin physical activity score)32. We confirmed that MLR assumptions were met (by plotting histograms of residuals and plotting residuals against predicted values). The relationship between arterial stiffness (AIX@75) and the 4 other health status measures (EuroQol, overall health, overall arthritis, damaged joint count) was also assessed by adjusting for the same 10 independent variables. Analysis was done using SPSS v. 17. Categorical variables were summarized using frequencies (percentages); normally distributed continuous variables are summarized as means (standard deviation, SD); skewed variables are summarized as medians (interquartile range, IQR). All participants provided informed written consent. The study was approved by Grampian Research Ethics Committee (study reference 04/S0801/67) and adhered to the Declaration of Helsinki.
RESULTS
We recruited 114 patients [mean age 54 yrs; 93 women (81%)] with a rheumatologist’s diagnosis of RA (83% with RF > 29 IU/ml) and median arthritis duration of 10 years (IQR 4–17 yrs). The majority were currently prescribed disease-modifying antirheumatic drugs (DMARD; 91%) for rheumatic disease and nonsteroidal antiinflammatory drug (NSAID) therapy (70%). Median ESR at assessment was 16 mm/h (IQR 8–28), with a median HAQ disability score of 1.13 (IQR 0.50–1.75; Table 1). Only 56% of patients met 4/7 ACR criteria for RA. Women had a longer duration of arthritis than men, a higher ESR at assessment, a higher HAQ score, and were somewhat younger (Table 1). Mean fasting cholesterol was 5.3 mmol/l, BP was 125/82 mm Hg (18% on antihypertensive drugs), and 24% were current smokers (41% never smoked; 35% ex-smokers). Two patients had diabetes (identified by fasting glucose; no one else with diabetes) and 2 patients were prescribed a statin. Other CV drugs (British National Formulary) included bendroflumethazide (n = 16), atenolol (n = 11), angiotensin-converting enzyme inhibitors (n = 5), and calcium channel blockers (n = 4). Mean augmentation index (AIX@75) was 31.5 (SD 7.7) and was higher in women than men. Women had a lower BP than men (124/81 vs 129/83 mm Hg); were less likely to be a current smoker (20% vs 38%); and were more likely to eat fruit and vegetables daily. Although they were more physically active overall than men (Table 1), women were less likely to engage in vigorous physical activity sufficient to work up a sweat (38% vs 43%, respectively).
A higher HAQ score and damaged joint count indicate a higher level of disease severity, while a higher EuroQol score and overall health/arthritis score (based on a 100-point scale) indicate a higher quality of life. As would be anticipated, the HAQ was positively correlated with the damaged joint count and negatively correlated with the EuroQol, health-overall, and arthritis-overall scores (Table 2). The highest correlation (Spearman’s ρ −0.73) was between the disease-specific (HAQ) and the generic (EuroQol) health-related quality of life (HRQOL) instruments. Women and men had similar EuroQol, health-overall, and arthritis-overall scores (a higher score indicating better health on a scale up to 100). The severity of arthritis was somewhat higher among women than men when assessed by damaged joint count and HAQ (Table 2).
The HAQ was positively correlated with arterial stiffness (AIX@75), with a Pearson correlation coefficient r of 0.42 (95% CI 0.25–0.56; Figure 1). On unadjusted analysis, a 1-point increase in the HAQ was associated with a 4.2-point (95% CI 2.5–6.0) increase in AIX@75. The HAQ alone explained 17% (adjusted R2) of the variability in arterial stiffness. On multivariable analysis, a 1-point increase in HAQ was associated with a 2.8 increase (95% CI 1.1–4.4; p = 0.001) in AIX@75. The regression model was highly statistically significant and explained almost half of the variability (adjusted R2) in arterial stiffness (Table 3). Based on a comparison of standardized regression coefficients, female sex, mean arterial BP, and HAQ demonstrated the strongest relationship with arterial stiffness (with coefficients of 0.38, 0.33, and 0.27, respectively; p ≤ 0.001). The limited influence of age in the regression model (standardized regression coefficient 0.04, p = 0.6) probably relates to the restriction of our study to adults aged 40–65 years.
Mean AIX (unadjusted for heart rate) was 34.8 (SD 8.3). AIX for men was 29.0 (SD 9.4), and for women 36.1 (SD 7.4). HAQ was correlated with AIX (r = +0.29; 95% CI 0.11–0.45), and on unadjusted analysis a 1-point increase in HAQ was associated with a 3.2 increase (95% CI 1.2–5.1) in AIX (although the HAQ alone explained only 8% of the variability in AIX). On multivariable analysis (with the same 10 variables as in Table 3), a 1-point increase in HAQ was associated with a 2.5 increase (95% CI 0.3–4.7, p = 0.03) in AIX. While the multivariable model was highly statistically significant (p = 0.0003), it explained only 20% (adjusted R2) of the variability in AIX.
The size of our study restricted the number of independent variables that could be included in the multivariable analysis. Several factors measured in the study are not included in the multivariable model reported in Table 3. The addition of individual cardiovascular variables (body mass index, fasting glucose, family history of coronary artery disease, CV drug therapy, treated hypertension) and rheumatological variables (RF-seropositive, current therapy with DMARD, NSAID, or prednisolone) made no appreciable difference in the HAQ regression coefficient, which ranged from 3.09 (with the inclusion of NSAID therapy) to 2.58 (with the inclusion of Carstairs deprivation score). The inclusion of additional variables did not improve the overall goodness to fit of the regression model.
The HAQ score was significantly correlated (p < 0.001, nonparametric Spearman’s ρ) with the other 4 health status measures. The strongest was a negative correlation with the EuroQol (ρ −0.73; Table 2). On multivariable analysis (adjusting for the same 10 independent variables), a 10-point increase in the EuroQol score was associated with a nonsignificant −0.36 reduction (95% CI −0.85 to +0.14; p = 0.16) in arterial stiffness (Table 3). The number of damaged joints was positively correlated with the HAQ (ρ +0.53), and on multivariable analysis each additional damaged joint was associated with a statistically significant 0.17-point increase (95% CI 0.04–0.29; p = 0.009) in arterial stiffness.
Overall arthritis and overall health scores (on a 100-point VAS) were negatively correlated with HAQ scores (ρ −0.61 and −0.43, respectively). On fully adjusted analysis, a 10-point increase in overall-arthritis score was associated with a statistically significant −0.78 reduction (95% CI −1.35 to −0.20; p = 0.008) in arterial stiffness. A 10-point increase in overall health score was associated with a −0.82 reduction (95% CI −1.65 to +0.01; p = 0.05) in arterial stiffness, but this was not statistically significant.
The regression models for these 4 health status measures explained a similar amount of the variation in arterial stiffness (adjusted R2) as the regression model including the HAQ, but only the arthritis-based health status measures (HAQ, damaged joint count, and overall arthritis) demonstrated a statistically significant relationship with arterial stiffness (Table 3).
DISCUSSION
We have demonstrated that in patients with RA, the level of arterial stiffness is significantly correlated with several health status measures. A 1-point increase in HAQ disability is associated with a statistically significant 3-point increase in arterial stiffness, independent of RA characteristics, major cardiovascular risk factors, and physical inactivity. Such an increase in arterial stiffness is equivalent to 6 years of normal arterial aging in healthy women between 45 and 55 years34. It is also comparable with the reduction in arterial stiffness observed in patients with RA after 6 weeks of treatment with atorvastatin17. A 1-point difference in HAQ between patients with RA equates with a 74% increased relative risk of death over 10 years9.
In our study, a research nurse obtained high-quality PWA measurements in optimal circumstances. We had few missing data and were able to include all participants in the multivariable analysis. We included patients with CV risk factors (such as smoking, hypercholesterolemia, and hypertension, adjusting for these in the analysis), but carefully excluded those with overt arterial disease (and limited age to 40–65 yrs) to avoid ceiling/flooring effects in arterial stiffness. Since AIX varies with heart rate, the main analysis used AIX standardized to a heart rate of 75 bpm (AIX@75)28. Resting heart rate is an independent risk factor for cardiovascular events (a faster heart rate is associated with increased risk)35. Consequently, heart rate may be a marker for subclinical atherosclerosis. We recruited participants from a consecutive series of patients attending routine rheumatology clinics and the characteristics of patients were comparable to those attending outpatient clinics elsewhere in the UK36.
Our cross-sectional study has some important limitations. Cardiovascular risk factors measured on a single occasion (such as BP and cholesterol) may not always accurately reflect previous levels. The prevalence of hypertension in our study (18%) is lower than described elsewhere37 and probably relates to the exclusion of patients with overt arterial disease (who would be anticipated to have higher levels of cardiovascular risk factors such as hypertension). We included ACR criteria as a variable in the regression analysis because of concerns that our study population might be clinically heterogeneous. While all participants had a rheumatologist’s diagnosis of RA, only 56% met ACR criteria for RA33. This may relate to a typographical error in our patient questionnaire, which asked about morning stiffness for 6 months (rather than 6 weeks). ACR criteria also “accumulate” over time38. In our study, 66% (36/55) of patients with a duration of arthritis greater than 10 years met ACR criteria. Pulse wave velocity was not measured and the relatively small size of our study restricted the number of independent variables that could be included in our multivariable analysis. Finally, recruitment of patients attending hospital clinics may have excluded patients with quiescent, mild, or burned-out disease managed solely in primary care.
Despite a careful search of the literature, we have been unable to identify studies that reported in detail on the relationship between arterial stiffness and HAQ disability (or other health status measures) in patients with RA. Several studies measured AIX in patients with RA using the SphygmoCor device, but none adequately explored the relationship between arterial stiffness and health status10–18. Only 1 study included the HAQ and reported a weak correlation (Spearman’s ρ = 0.19) with arterial stiffness10, which is lower than the correlation (Spearman’s ρ = 0.42) observed in our study. Two studies reported including a 100-point patient-derived current disease activity score (DAS)12 and a global assessment score15, but neither study reported any results relating to these scores. Three studies included the 28-joint count DAS28 composite and reported on the correlation of DAS28 with the AIX (coefficients of 0.13, 0.21, and 0.24)10,12,15.
Three prospective cohort studies, each of more than 1000 patients with RA, have shown the HAQ to be predictive of overall mortality, although none reported on cardiovascular mortality7–9. A study of 1384 patients with RA followed over 10 years in the United States reported an adjusted hazard ratio of 1.74 (95% CI 1.43–2.11) for all-cause mortality for each 1-point increase in HAQ score7. The Norfolk Arthritis Register, a community-based UK inception cohort study of 1098 patients with inflammatory polyarthritis, reported an adjusted OR for death of 1.29 (95% CI 1.03–1.62) for patients with a HAQ ≥ 1.5 at baseline after 10 years of followup8. A study of 1387 patients with RA over 20 years in the state of Kansas in the US (7339 person-yrs followup) also found the HAQ to be a strong predictor of all-cause mortality (adjusted OR 2.0, 95% CI 1.5–2.5, for a 1 SD change in HAQ)39. In the same study, patient self-report of global disease severity on a 10-point VAS was also a strong predictor (unadjusted OR 1.8 for a 1 SD change in VAS; p < 0.001, 95% CI not reported). Both the HAQ and global disease severity were stronger predictors of death than either laboratory or radiographic measurements39.
Ours is the first study to confirm that RA-related disability assessed using the HAQ is significantly and independently correlated with arterial stiffness in patients with RA who do not have preexisting arterial disease. Research has shown that the HAQ is a strong independent predictor of all-cause mortality. Further research is required to establish whether the HAQ is a strong independent predictor of CV mortality and whether the HAQ should be incorporated into the clinical prediction of CV outcomes in patients with RA.
Acknowledgments
We are grateful to Dr. John Meecham, MD, FRCP (for ECG reporting), Dr. David Crosbie, FRCP, BSc (for developing the initial study protocol and obtaining ethical approval), and to all the patients who participated in our study.
Footnotes
-
Supported by the National Health Service Grampian Rheumatology Endowments.
- Accepted for publication December 4, 2009.