There is considerable interest in the association between inflammatory arthritis and cardiovascular disease, but evidence from primary care settings emphasizing this association is sparse. To the best of our knowledge, this is the first study to report on the CVD prevalence rate in inflammatory arthritis, diabetes mellitus and osteoarthritis patients compared with controls in a nationally representative primary care population. We report nearly identical age, sex, hypertension and hypercholesterolemia adjusted CVD prevalence rates in patients with either inflammatory arthritis or diabetes mellitus, supporting our previous observations that the magnitude of CVD burden in inflammatory arthritis is comparable with that in diabetes mellitus [9
The fact that adjustment for hypertension and hypercholesterolemia attenuated the association between CVD and diabetes mellitus, but far less so the association between CVD and inflammatory arthritis, is another important observation. Although it is not possible to report cause-and-effect relationships with the cross-sectional design, there are two possible explanations for this observation. First, given the lower prevalence rate of hypertension and hypercholesterolemia and lower use of statins and anti-hypertensive agents in inflammatory arthritis patients compared with diabetes mellitus, one may speculate that classic risk factors are underreported and hence undertreated in inflammatory arthritis. CVD risk management is one of the main goals in the treatment of diabetes mellitus, but, at least until recently, not in inflammatory arthritis patients, which supports this hypothesis. Second, factors other than classic CVD risk factors potentially explain the strong association between inflammatory arthritis and CVD. Indeed, it is increasingly hypothesized that individuals with inflammatory arthritis are more affected by CVD because of a chronic systemic inflammatory process, as this accelerates the development of all stages of atherosclerosis [17
]. Inflammatory cells are commonly found within atherosclerotic lesions, particularly in the culprit lesions, and many inflammatory cells are implicated in the pathogenesis of atherosclerosis from its earliest stages [6
]. Additionally, inflammation may indirectly increase CVD risk through deteriorating CVD risk factors [18
]. The fact that osteoarthritis (a non-systemic inflammatory comparator) did not convey any excess CVD risk after adjustment for CVD risk factors is another important observation and may comply with the inflammatory hypothesis. Surprisingly, after adjustment, osteoarthritis was negatively associated with CVD. A possible explanation for this finding is selective drop-out (‘healthy survivors effect’) in the group osteoarthritis patients. Because of the high mean age of this group (about 70
years old) it is likely that the patients with the highest risks already died, resulting in an underrepresentation of the prevalence rate of CVD.
Using data from electronic medical records from general practices makes it possible to compare the prevalence of CVD between different patient groups in the same study population. This has the advantage that all morbidity is measured in the same manner, minimizing selection bias. Another strength of this study is the use of a representative group of patients with a mean average disease severity. The fact that the observed CVD prevalence rates in inflammatory arthritis and diabetes mellitus concur with other studies lends external validity to our observations [7
]. Also, of important note is the adjustment for important CVD risk factors (age, gender, hypertension, and hypercholesterolemia) on the one hand, and controlling for amplification of CVD risk by individuals with more than one disease on the other hand. Some other studies also reported an increased prevalence of CVD in inflammatory arthritis from a primary care population, but neither of these studies controlled for hypertension and/or hypercholesterolemia [7
Besides the advantage of having access to large representative patient populations, using data from health registries also have some disadvantages. First, the large number of patients makes it impossible to validate diagnoses and to collect additional data in terms of time and funding. Although diagnoses are not validated by a medical specialist, the used GP diagnoses are based on the guidelines of the Dutch College of General Practitioners (NHG). Furthermore, medical specialists send a letter to the GP after the consultation of a patient with information about diagnosis and treatment. It is likely that the GP uses this information to register the diagnosis inflammatory arthritis in the EMR of the patient. Therefore, we expect an overestimation of the number of IA patients in our primary care cohort, which might resulted in an underestimation of our results. Second, data in health registries are not collected in a structural way and patients are not measured periodically. Data are only available from patients who visit their GP for a health problem. As a results, important confounders, such as smoking and obesity, are not always registered and these variables cannot be used in the statistical analyses. Also, not all prescribed drugs in specialised care are registered in the EMRs of GPs, which makes it not possible to measure the use of DMARDs reliably. This hampers us to adjust for these potentially important confounders (residual confounding). A limitation of the cross-sectional design is a lack of information on the sequence of the disease and CVD events, and hence a causal relationship between the advent of inflammatory arthritis and development of CV disease can not be proven.