Patients
Demographic, RA and CVD related data were collected from 613 RA patients by chart review. This sample of RA patients was randomly taken from the entire RA population registered in the Jan van Breemen Institute, a large rheumatology outpatient clinic in Amsterdam, the Netherlands. All patients fulfilled the American College of Rheumatology criteria of RA. Patients were recruited from time of diagnosis, between 1953 and 2002, onwards until March 2004, the end of the follow up period.
Study design
A case control study of incident CVD was performed, comparing 72 patients with RA and CVD to 541 RA patients without CVD. CVD was evaluated from the time of diagnosis of the RA up to the time of the first cardiovascular event or until the end of the follow up period.
Cardiovascular disease and risk factors for it
CVD was defined as a verified medical history of coronary, cerebral or peripheral arterial disease. Coronary artery disease included a history of myocardial infarction, a coronary artery by-pass graft procedure, a percutaneous transluminal coronary angioplasty or ischemic abnormalities on ECG. Cerebral arterial disease was defined as a history of cerebral vascular accident (confirmed by a neurologist), a transient ischemic attack or a carotid endarterectomy. Peripheral arterial disease included an aneurysm of the thoracic and/or abdominalis aorta, a peripheral arterial by-pass operation and amputation of the (lower) leg. Assessed risk factors for CVD were age, male sex, hypertension, diabetes, hypercholesterolemia, and smoking habits. Hypertension, diabetes and hypercholesterolemia were considered to be present if patients received treatment for these conditions. Smoking habits were recorded as use ever versus never. All these variables were monitored throughout the entire disease duration.
Statistical analyses
Comparisons between the various DMARD groups and between the RA patients with CVD and without CVD were performed using Students' t-tests and Mann-Whitney U-tests for continuous variables and Pearson's Chi-square tests for dichotomic variables.
The dataset was categorized into groups according to the use ever of sulfasalazine (SSZ), hydroxychloroquine (HCQ) or MTX, either as monotherapy or as combinations of these drugs (both sequentially and concurrently in time). The final group consisted of patients who never used any of the three major DMARDs; this resulted in a total of eight groups. These groups were chosen because SSZ, HCQ and MTX are the most commonly used drugs and well represented in our random sample of RA patients.
Logistic regression modeling was used to calculate the odds ratios (ORs) and 95% confidence intervals (95% CIs) of CVD for the various DMARD groups simultaneously. In the regression analysis the group of RA patients who never used SSZ, HCQ or MTX was used as the reference group with a preset OR of 1.00.
The first regression model corrected for age, gender, smoking ever and RA duration. Correcting for age, gender and smoking was done because these variables are known to be associated with CVD but not with the use of certain DMARDs. Correcting for RA duration was done because the chance for a patient to be treated with more than one DMARD increases the longer the duration of the disease. As an additional analysis prednisone use ever was added to this first model.
In the second regression analysis we added the presence of hypertension, diabetes and hypercholesterolemia to the first model. Adding these known risk factors for CVD was done for two reasons. Firstly, because these risk factors could be over- or under-represented in certain DMARD groups and, therefore, falsely influence the cardiovascular risk for these groups. Secondly, correction for known cardiovascular risk factors was done to explore possible pathways by which the investigated DMARDs can influence cardiovascular risk; for example, a DMARD could increase the cardiovascular risk by causing hypertension and this increased risk would disappear after correcting for hypertension.
A third analysis was done using the first model and adding the presence or absence of a positive rheumatoid factor test and erosions on radiographs. This enabled us to calculate the ORs for CVD associated with these two RA related variables.
The three models described above were also used to explore if there was any dose dependency in the possible associations between the DMARD groups and CVD risk. Therefore, we determined the presence of interactions between any of the DMARD groups and the maximum used dosages, days of DMARD use and a cumulative variable. Because the maximum dosages of the different DMARDs are of different quantities (for example, 30 mg for MTX and 3,000 mg for SSZ) we calculated the percentage of the maximum dosages allowed by the Dutch and European medication agencies to be prescribed. For example, 30 mg is the maximum dosage allowed to be prescribed for MTX; therefore, if a patient only used 15 mg at the most, we used 50% as the maximum dosage in the calculations. This way we were able to compare the various DMARD dosages. As a cumulative variable we calculated the maximum percentage of the highest prescribable dosage multiplied by the years this DMARD was used.
A p value of 0.05 or smaller was considered statically significant and all tests were performed using the SPSS 12.0 software package for Windows (SPSS Inc., Chicago, IL, USA).