Patients were assessed according to a standard protocol to evaluate rheumatoid arthritis (SPERA) 13. The rheumatologists performed a clinical assessment including swollen and tender joint counts. Information concerning extra-articular features and comorbidities, including CV events, was established by a record review, a detailed clinical examination, and asking the patient at the time of the visit. The use of all DMARDs, including dates of start and discontinuation of each DMARD, was recorded. The most recent rheumatoid factor (RF) values were collected; RF was considered positive or negative according to the local reference values any time over the course of the disease. No training as to how to collect data or to perform joint counts was provided, and the study was intended to reflect routine clinical practice. All patients had fulfilled the American College of Rheumatology 1987 criteria for the classification of RA during the course of the disease 14.

The presence of subcutaneous nodules, lung disease (nodules, fibrosis, or pleuritis), Felty syndrome, vasculitis, pericarditis, and scleritis was counted for extra-articular disease. CV events included myocardial infarction, angina, coronary disease, coronary bypass surgery, and stroke. Dates of these events were recorded. These data were based on participating rheumatologists' reports of CV events on their patients and no other confirmation was required. The presence of hyperlipidemia, diabetes mellitus, and hypertension was recorded. No type specifications for hyperlipidemia or diabetes mellitus were made.

The patients completed an expanded self-report health questionnaire that was first translated to each language; the study is explained elsewhere in more detail 12. The questionnaire included the Health Assessment Questionnaire (HAQ) 15, years of education, height and weight for body mass index (BMI), and lifestyle choices such as smoking and frequency of physical exercise. Smokers were classified as 'nonsmokers' and 'ever-smokers', including 'current smokers'. Obesity was defined as BMI of greater than 30. 'Physical inactivity' included patient responses 'no exercise' or '1 to 2 times a month' versus exercises 'regular exercise one or more times a week'.

Data are presented as means with standard deviations (SDs) and percentages with 95% confidence intervals (CIs). The Student t test and chi-square test were used for comparison between groups.

The prevalence of CV events was calculated including all patients in the cohort who reported to have had a CV event. Patients with a CV event before the diagnosis of RA were excluded from the risk factor analyses. Time-to-event was calculated from the date of diagnosis of RA; patients with no CV event were censored at the date of evaluation. Univariate Cox proportional hazard regression models were computed to estimate the risk of any CV event including each of the traditional CV risk factors and disease characteristics as independent variables in separate models. Second, multivariate Cox proportional hazard regression models were computed to estimate the risk of all CV events, myocardial infarction, and stroke. Variables that were included in these models were age, gender, presence of RF, extra-articular disease, hypertension, hyperlipidemia, diabetes, smoking, obesity, physical inactivity, and country. Multivariate analyses were performed including all patients who had a CV event after a diagnosis of RA and separately for 'high' and 'low' CV prevalence countries (that is, countries above and below the median, respectively).

Time of exposure to each DMARD was calculated as the time (in years) that elapsed between the date of start of a DMARD and the date of discontinuation, CV event, or date of evaluation, which ever happened first. Each DMARD was analyzed independently in a Cox regression model, first unadjusted and then adjusted for age, gender, disease activity/severity (DAS 28 [disease activity score using 28 joint counts] and HAQ), RA characteristics (RF positivity and presence of extra-articular manifestations), and the presence of traditional CV risk factors (hypertension, hyperlipidemia, diabetes, smoking, and obesity).

Data collection began in January 2005, and in October 2006 the QUEST-RA project included 4,363 patients at 48 sites in 15 countries. Concerning demographic variables, this cohort represents a typical RA population: 78% of patients were female, 90% were Caucasian, the mean (SD) age was 57 (14) years, the mean disease duration was 11 (9) years, and the mean duration of education was 10 (4) years. Overall, 74% of patients had positive RF, ranging from 61% in Germany to 90% in Argentina. Extra-articular disease was present in 24% of patients, ranging from less than 15% in Italy and the Netherlands to 34% in Denmark (Table 1).

The overall prevalence of CV morbidity (myocardial infarction, angina, coronary disease, or stroke) was 9.3%, with considerable variation between countries (less than 5% in Argentina and France and greater than 10% in Finland, Germany, Poland, the UK, and the USA) (Table 1). CV events were more prevalent in men than in women (Table 2). The overall prevalence for the whole cohort of lifetime myocardial infarction was 3.2%, and the prevalence for stroke was 1.9%.

The prevalence of CV risk factors was 33% for hypertension, 14% for hyperlipidemia, 8% for diabetes, 43% for smoking ever, 72% for physical inactivity, and 18% for obesity. Diabetes was more frequent in men (10%) than in women (7%) (Table 2). More men than women had ever smoked (68% versus 37%; P = 0.0001) or were current smokers (26% versus 15%; P = 0.0001).

In a univariate Cox regression analysis, extra-articular disease was statistically significantly associated with CV morbidity (Table 3). Among traditional CV risk factors, age, gender, hypertension, hyperlipidemia, ever-smoking, and diabetes showed a statistically significant association with CV events.

In a multivariate Cox regression analysis, older age, male gender, hypertension, hyperlipidemia, and ever-smoking were independently associated with occurrence of CV events (Table 4). Extra-articular RA (hazard ratio [HR] 2.26; 95% CI 1.29 to 3.97), hyperlipidemia (HR 3.51; 95% CI 1.98 to 6.21), and ever-smoking (HR 3.20; 95% CI 1.74 to 5.90) were all associated with myocardial infarction although association of extra-articular disease was not statistically significant in 'low prevalence' (of CV disease) countries (Table 4). Hypertension (HR 2.81; 95% CI 1.49 to 5.30) and diabetes (HR 2.23; 95% CI 1.12 to 4.44) were associated with stroke; associations were statistically significant for hypertension in 'high prevalence' countries and for diabetes in 'low prevalence' countries (Table 4).

Patients who had hypertension were treated less frequently with methotrexate and biologic agents compared with patients who had no hypertension, but the former were treated more frequently with leflunomide and glucocorticoids (data not shown). Table 5 shows the HR for the occurrence of CV events by year of exposure to each DMARD, when adjusted to age, gender, disease activity, and traditional risk factors. One year of methotrexate use was associated with 15%, 18%, and 11% decreases of risk for all CV events, myocardial infarction, and stroke, respectively. Leflunomide was also associated with a reduced risk of CV events, and glucocorticoids and sulfasalazine were associated with a small but significantly reduced risk of all CV events. A lower risk for all CV events and myocardial infarction was also associated with a longer exposition-duration to TNF-α blockers (HR 0.42; 95% CI 0.21 to 0.81; P < 0.05).

Patients with RA are 30% to 60% more likely to suffer a CV event compared with the general population 1617, especially myocardial infarction 181920, whereas the incidence and prevalence of stroke generally have been reported to be similar in RA as in the general population or in patients with osteoarthritis 1820. Only one study found higher prevalence of stroke in RA than in controls 16. We found a lower prevalence of stroke compared with other cross-sectional studies 1618, although comparative data for the reference from general populations were not available in our study.

The crude prevalence of CV events differed between countries (Table 1). Results in the present study, to some extent, are similar to those in the World Health Organization MONICA (Multinational

Moni

Ca

In our study, the frequency of CV events was double in men compared with women (four times for myocardial infarction) and does not provide any surprises compared with observations in the general population 21. In univariate analyses, all traditional CV risk factors, except obesity and physical inactivity, were associated with CV morbidity, and in multivariate models, hypertension, hyperlipidemia, diabetes, and ever-smoking remained independent risk factors. Thus, our results confirm the role of traditional risk factors concerning CV morbidity in patients with RA. However, it needs to be recognized that our analyses are restricted to people who survived following a CV event due to the cross-sectional nature of the study and some of the predictors being identified may be predictors of survival following a CV event rather than the occurrence of the event. Furthermore, we did not collect data on family history of CV disease, which appears a significant risk factor for CV morbidity also in patients with RA 23.

Physical inactivity is common in the general population and a frequent consequence of arthritis. In the general population, the frequency of weekly physical activity of three or more times is associated with reduced CV morbidity 24. In the present study, only 13% of patients exercised three or more times a week 25. Physical inactivity (exercise frequency '1 to 2 times a month' and 'none') was not associated with CV disease in our analyses.

RA is associated with premature and accelerated atherosclerosis 1626. Carotid artery ultrasound and coronary angiography have shown that atheromatosis is more frequent in RA than controls 2728 and is related to the extent of radiographic damage of joints, which, in turn, is indicative of sustained inflammation 29. Moreover, multivessel coronary artery calcifications in RA are related to smoking and an elevated sedimentation rate 4. An elevated C-reactive protein at baseline predicted CV mortality over the follow-up of 10 years in patients with inflammatory polyarthritis in general and particularly in RF-positive patients 3031. Some processes intrinsic to the pathogenesis of RA play important roles in CV damage and its clinical consequences: abnormal endothelial function 32, autoantibodies against oxidized low-density lipoprotein 33, and levels of serum mannose-binding leptin 34.

Van Halm and colleagues 9 found an association between CV disease and seropositive and erosive RA, and Turesson and colleagues 835 between CV morbidity and mortality and extra-articular disease. The present study lends further support to the significant association between extra-articular disease and CV morbidity, and this observation underlines higher CV morbidity in severe RA.

Some of the medications used to treat RA might increase the risk for CV morbidity. The risk of myocardial infarction is increased in nonsteroidal anti-inflammatory drug (NSAID) and cyclooxygenase 2 inhibitor users, especially with rofecoxib 3637.

Glucocorticoids might increase CV morbidity by promoting proatherosclerotic lipid profiles 3839. Two recent articles show an increased rate of CV events with long-term use of glucocorticoids 1040. However, Davis and colleagues 41 found no association between cumulative glucocorticoid exposure and CV events in RA patients followed for a median of 15 years, after adjusting for other CV risk factors and markers of RA activity. In the present study, long-term use of glucocorticoids was independently associated with a reduced risk of 'all' CV events (Table 5).

The use of methotrexate has been associated with a significantly lower risk for CV events in RA patients compared with patients who had never used DMARDs 9. Long-term follow-up of RA patients has shown that the use of methotrexate is significantly associated with reduced overall and CV mortality 42.

Suissa and colleagues 10, in a case control study, found that the use of DMARDs was associated with a decreased rate of myocardial infarction. However, Solomon and colleagues 40, in a case control study, found a higher risk for CV events in patients who had received cyclosporine, azathioprine, or leflunomide. In our study, longer use of DMARDs such as methotrexate, leflunomide, and sulfasalazine was associated with a decreased risk of CV disease. The association remained statistically significant for myocardial infarction with methotrexate (P < 0.001) and sulfasalazine (P < 0.05), even when adjusted for disease severity and traditional risk factors such as smoking, which is a risk factor for both CV disease and RA 43. In contrast, the use of antimalarials and intramuscular gold was not associated with a change of the risk of CV disease. In one report, atherogenic lipid profiles of RA patients improved after specific therapy for arthritis, primarily due to the increase of serum high-density lipoprotein cholesterol levels 44.

With respect to TNF-α blockers, infliximab may improve endothelial function in RA after 12 weeks of therapy 45, suggesting that inflammation is a mediator of endothelial dysfunction, although such beneficial effects do not appear to sustain for a long time 4647. A recent study suggested that the risk for developing first CV events in RA is lower in patients treated with TNF-α blockers 11. However, that study did not control for most of the traditional risk factors for accelerated atherosclerosis. On the other hand, two case control studies showed no reduction of infarction rate in RA with TNF-α blockers 1041. Other recent reports from large databases show discordant results on CV disease incidence in TNF users versus nonusers 484950. In the present study, longer use of TNF-α blockers was associated with a reduced risk of CV disease although limited availability of biologics might interfere with the results. Furthermore, patients with suspected CV disease may not be prescribed biologic agents.

Several limitations are to be recognized. First, QUEST-RA is a cross-sectional study with a possible left censorship concerning CV events. Many patients may have experienced a fatal CV event and, therefore, could not be included in this study. Second, all data are based on participating rheumatologists' reports and no verification of data was performed. Therefore, it is possible that some data could be missed. Third, data on the longitudinal use of NSAIDs or cyclooxygenase 2 inhibitors were not collected. Fourth, the extent of radiographic erosions (representing cumulative disease activity) was not analyzed in the present study. Despite limitations, QUEST-RA is a unique program that has succeeded to date in collecting data on clinical RA patients according to an identical protocol in various locales in various countries and various cultures and provides data that are not available from any other resources at this time.