The present study is the first nationwide study of IMDs and ischemic and hemorrhagic stroke. The results indicate that several IMDs increase the risk of hospitalization for both ischemic and/or hemorrhagic stroke. The relative risk of ischemic and hemorrhagic stroke during the first year after hospitalization with certain IMDs was even higher than the risks associated with many traditional risk factors for ischemic and hemorrhagic stroke [1, 25]. Although it declined over time, the overall risk of ischemic and hemorrhagic stroke remained elevated for 10 or more years for some IMDs. The results of our study are in line with previous studies linking rheumatoid arthritis [3, 5, 6, 8–12, 20], systemic lupus erythematosus [3, 5, 6, 8, 13–15, 19], Sjögren´s disease [3, 5, 6, 16], systemic vasculitis [3, 5], inflammatory bowel disease [3, 5, 8, 17], and psoriasis [8, 18] to an increased risk of cardiovascular disease. However, what distinguishes our study from these other studies is its comparison of large numbers of patients and 32 types of IMDs with the general population in a nationwide setting, as well as the long-term follow-up of patients and the determination of risk for both ischemic and hemorrhagic stroke. Moreover, we also found a number of novel associations between IMDs and ischemic and hemorrhagic stroke. The results of the present study suggest that increased risk of subsequent ischemic and hemorrhagic stroke is a common feature of several IMDs, not just selected conditions such as systemic lupus erythematosus  and rheumatoid arthritis .
Although the increased risk of ischemic and hemorrhagic stroke may have different underlying causes in different IMDs, a general link between systemic inflammation and atherothrombosis has been indicated [2–8]. In some conditions, such as in immune thrombocytopenic purpura, hemorrhagic stroke may occur as the direct result of thrombocytopenia. The formation of autoantibodies may, in special cases, also contribute to stroke . The increased risk of stroke may be specific for more severe cases of IMDs, since the patients in our study had been admitted to hospital. The effects of treatment—corticosteroids promote hemostasis —and the effect of inflammation on coagulation  may also contribute to the identified associations. Hypothetically, the fact that the risk of ischemic and hemorrhagic stroke decreased over time may suggest that it is linked to the inflammatory activity of the IMDs, which is likely to decrease over time due to treatment. In line with this hypothesis, in several studies disease activity appears to be linked with atherosclerosis progression [2–8, 28, 29]. However, as we lack treatment data, we cannot prove this hypothesis but in this context it is interesting that the relative risk of both hemorrhagic and ischemic stroke was lower between 1997 and 2008 than between 1987 and 1996 (Additional file 1 Tables S12 and S13).
The present study has certain limitations. For example, we had no data on general cardiovascular disease risk factors such as weight, smoking, and diet. It is unrealistic to gather such data for an entire national population. However, we did adjust for socioeconomic status, which is associated with risk factors such as smoking. Aspirin and non-steroidal anti-inflammatory drugs (NSAID) may affect the risk of ischemic and hemorrhagic stroke [30, 31]. However, we had no access to treatment data. Adjustment was, however, made for several comorbidities (chronic lower respiratory diseases, obesity, alcoholism and alcohol-related liver disease, type 2 diabetes mellitus, hypertension, atrial fibrillation, coronary heart diseases, heart failure, renal disease and sepsis). Still, residual bias may remain due to hospitalization of the most severe cases with IMD. However, all cases with previous or coexisting stroke were excluded to avoid selection bias. Totally, 8113 IMD patients with previous or coexisting ischemic stroke and 1416 with hemorrhagic stroke were excluded from the study, which in turn instead may underestimate the stroke risk. In fact, our results are within the limit for published cardiovascular disease risk in IMDs like RA [3, 5, 6, 8–12, 20] and SLE [3, 5, 6, 8, 13–15, 19]. Thus, the estimated risks of stroke in IMD patients appear to be fairly valid. Anyway, the present study reflects the real world risks for stroke among hospitalized IMD (without previous stroke or at the same time as first hospitalization for IMD). All cases of ischemic and hemorrhagic stroke in Sweden should, according to official guidelines, be treated at hospitals . Moreover, hospitalization incidence rates were calculated for the whole follow-up period, divided into five time periods, and adjustments were made for possible changes in hospitalization rates over time.
This study also has a number of strengths. The study reflects the situation in real world medicine during 22 years in a country with a high standard in the medical diagnosis [22, 23, 33–35]. The study population included all individuals clinically diagnosed with IMD and ischemic and hemorrhagic stroke in hospital during the study period, which eliminated recall bias. Because of the personal identification number assigned to each resident in Sweden, it was possible to trace all subjects for the whole follow-up period. Data on occupation were 99·2% complete (1980 and 1990 censuses), which enabled us to adjust our models for socioeconomic status. A further strength of the present study was the use of validated hospital discharge data. The Hospital Discharge Register has high validity [22, 23, 33–35], especially for cardiovascular disorders such as stroke, for which approximately 95% of diagnoses have been shown to be correct [33–35]. Though, the positive predictive value (PPV) may differ between diagnoses in the Swedish Hospital Discharge Register, the PPV is generally around 85-95% .