Risk of Stroke in Patients with Rheumatism: A Nationwide Longitudinal Population-based Study

The aim of this study was to investigate rheumatoid arthritis (RA), and systemic lupus erythematous (SLE) as risk factors for stroke. The study was analyzed by Using the Taiwan Longitudinal Health Insurance Database 2005 (LHID2005), this cohort study investigated patients with a recorded diagnosis of RA (N = 6114), and SLE (N = 621) between January 1, 2004, and December 31, 2007, with age-matched controls (1:4) (for RA, N = 24456; SLE, N = 2484). We used Cox proportional-hazard regressions to evaluate the hazard ratios (HRs) after adjusting confounding factors. Our study found 383 of 6114 RA patients, experienced stroke during the 20267 person-year follow-up period. The adjusted HR of stroke for RA patients was 1.24 (95% CI, 1.11 to 1.39), and for SLE patients was 1.88 (95% CI, 1.08 to 3.27). When steroid was added as additional confounding factor, the adjusted HR of ischemic stroke for RA patients was 1.32 (95% CI, 1.15 to 1.50), and for SLE patients was 1.31 (95% CI, 0.51 to 3.34). In conclusion, the rheumatic diseases of RA, and SLE are all risk factors for stroke. After controlled the effect of steroid prescription, RA is risk factor for ischemic stroke.

the SLE patients have higher risk of stroke had been investigated, the medication influence was not analyzed and further study is needed.
As mentioned above, studies have addressed the risk of stroke among patients with rheumatologic diseases. The relationship between rheumatologic disease and stroke has not yet been explored thoroughly. Studies that have investigated this relationship have revealed controversial results. Therefore, this study investigates rheumatologic diseases as risk factors for stroke with a nationwide population-based study.

Methods
Study population and study design. The Longitudinal Health Insurance Database 2005 (LHID2005), released by the Taiwan National Health Research Institutes, provided the database for our study. The database contains all the original claim data of 1000000 beneficiaries, including inpatient care, ambulatory care, dental care, prescription drugs, and ICD-9-CM (International Classification of Diseases, 9 th Revision, Clinical Modification) diagnostic codes. The beneficiaries were randomly sampled from 25.56 million persons in the Registry for Beneficiaries 2005,a system that covers almost 99% of the total population of Taiwan 19 . In the ethic aspect, the database used consisted of de-identified secondary data, the study met the requirements of the ''Personal Information Protection Act'' in Taiwan. The data were analyzed anonymously and the need for informed consent was waived approved by institution of review board.
The study cohort consisted of all patients who, according to the LHID2005, had been diagnosed autoimmune disease such as RA (ICD-9-CM codes 714.0), or systemic lupus erythematous (ICD-9-CM codes 710.0) between January 1, 2004, and December 31, 2007 (N 5 6508, RA; N 5 647, SLE). The ICD-9-CM was coded by a clinician specializing in rheumatology. The RA and SLE diagnoses are made according to diagnostic criteria stipulated by the American College of Rheumatology (ACR). For the purpose of a summarized diagnosis of RA and SLE, the 2 consecutive codings of RA and 15 consecutive SLE codings were incorporated in this study. Patients with missing variables such as date of birth and sex were excluded from the study (N 5 72, RA; N 5 18, SLE). Exclusion criteria included a diagnosis of stroke (ICD-9-CM codes 430-438) before that of RA, or SLE(N 5 322, RA; N 5 8, SLE).The resulting study cohort comprised 6114 RA, and 621 SLE patients.
We obtained control patients from the remaining patients in the LHID 2005 who had been registered between January 1, 2004, andDecember 31, 2007. Patients were excluded if they had been diagnosed as having RA, or SLE between 2004 and 2008 or with stroke before 2004. The selected control patients were matched with those in the study cohort (4 control patients per case patient) according to age (,530, 31-40, 41-50, 51-60, 61-70 and .70 y) and sex. All patients were observed from the date of cohort entry until they developed stroke or until the end of 2008.
Outcome measure. We identified the first diagnosis of stroke (ICD-9-CM codes 430-438) as the study end point. For stroke type analysis, we separated hemorrhagic stroke (ICD-9-CM codes 431) and compared the ischemic stroke in further adjusted hazard ratio analysis. All the study subjects were followed from the index date to occurrence of end point or until December 31, 2008, whichever was first, and the observations on the last dates were considered as censored observations. Statistical analysis. We compared demographic characteristics and co-medical disorders using Pearson's chi-squared test or Fisher's exact test. We used Cox models to evaluate hazard rates for stroke between the study cohort and control cohort after adjusting for potential confounding factors, including patient age (as continuous), sex, diabetes mellitus, coronary heart diseases, hypertension, hyperlipidemia, and urbanization level 20 . We checked each dichotomous variable in the model for proportionality by exploratory diagnostic log-log survival plots to meet the proportional-hazards assumption. We plotted the stroke hazard curves based on the Cox models for the patients and the control cohort after adjusting for potential confounding factors. For further analysis of medication related compounding factor analysis, steroid was taken as a variant for adjusted hazard ratio analysis among these two rheumatic diseases and different type of strokes. All analyses were performed using an SAS statistical package (SAS System for Windows, Version 9.1.3, SAS Institute Inc., Cary, NC, USA) and SPSS Version 20. A P value of ,.05 was considered statistically significant.

Results
We identified 6114 RA, and 621 SLE patients and age-and sexmatched control cohorts. Compared to the control cohort in Table 1, RA patients were more likely to have comorbid hypertension (P , .001), hyperlipidemia (P , .001),coronary heart disease (P , .001), congestive heart failure (P , .01), renal disease (P ,.001), and valvular heart disease (P , .001). SLE patients were more likely to have hypertension (P , .001), hypercoagulability disease (P , .01), renal disease (P , .001), and valvular heart disease (P , .05). Figure 1 shows the hazard curves for stroke in RA, and SLE patients and the control cohort during the 5-year follow-up period after adjustment for patient age, sex, diabetes mellitus, coronary heart disease, hypertension, hyperlipidemia, and urbanization level (log-rank test, P , .001). Table 2 shows the incidence and hazard ratios for stroke in the RA, and SLE patients and the control cohort. Of the 6114 RA patients, 383 (a rate of 18.89 per 100000 person-years) experienced stroke during the 20267 person-year follow-up period. After adjustment for potential confounding factors, the adjusted HR of stroke(with both ischemic and hemorrhagic types) was1.24 (95% CI, 1.11 to 1.39), which was larger than those in the control group. Of the 621 SLE patients, 21 (a rate of 8.18 per 100000 person-years) experienced stroke during the 2567 person-year follow-up period, and the adjusted HR of stroke was 1.88 (95% CI, 1.08 to 3.27) after adjustment for potential confounding factors. The adjusted HR of stroke for the entire study cohort was 1.25 (95% CI, 1.12 to 1.40). Table 3 shows the incidence and hazard ratios for ischemic stroke in the RA, and SLE patients and the control cohort. Furthermore, the steroid effect was taken as a confounding factor for ischemic stroke in this step of analysis. When steroid prescription was not enrolled in the confounding factor for ischemic stroke, all of the two rheumatic diseases are risk factors for ischemic stroke (adjusted HR: RA 5 1.24,95% CI, 1.02 to 1.50, P , .05 and SLE 5 2.04,95% CI, 1.07 to 3.91, P , .05)However, when steroid prescription was enrolled as confounding factor, RA (adjusted HR 5 1.32, 95% CI, 1.15 to 1.50, P , .001) was risk factor for ischemic stroke. But there is no statistic significance of SLE as a risk factor for ischemic stroke. This may indicate that steroid could influence the developing of ischemic stroke for SLE patients. (Figure 2)When analyzing hemorrhagic type of stroke by adjusted HR analysis, There are no statistical significance of adjusted HR of RA(with steroid usage P 5 .600; without steroid usage P 5 .115), and SLE (with steroid usage P 5 .427; without steroid usage P 5 .994) patients.

Discussion
Our study revealed that patients with autoimmune diseases such as RA, and SLE were more vulnerable to cerebral vascular accidents (CVA). And the analysis the risk of ischemic and hemorrhagic stroke types among these two rheumatic diseases revealed that there is no statistical significance of adjusted HR in hemorrhagic stroke. This can be explained that the pathogenesis theory of ischemic stroke and hemorrhagic stroke is different.
The pathogenesis of ischemic stroke in patients with autoimmune diseases, such as RA, and SLE, is described below. First, atherosclerosis is an immune-mediated inflammatory process 21 . The systemic inflammation in the pathogenesis of autoimmune diseases interacts and accelerates vessel atherosclerosis 22 . This mechanism is consistent with our observation that the incidence of stroke increased in patients who had suffered from autoimmune diseases for longer periods 23 . Second, endothelial dysfunction is a key pathologic process that occurs in early atherosclerosis. Endothelial function is impaired in systemic autoimmune diseases 24,25 . Third, traditional cardiovascular risk factors are more prevalent in patients with systemic autoimmune diseases 25 . Finally, medication for autoimmune diseases might play a role in the increased risk of atherosclerosis. NSAIDs are associated with increased cardiovascular risk and stroke 26,27 . Corticosteroids have been shown to be associated with higher cardiovascular risk, which might be a result of weight gain, adverse alterations in lipids, insulin resistance, and diabetes 28 .
Compared to studies for stroke risk of RA patients by Sodergren et al (odds ratio 5 2.6) and Semb et al (odds ratio 5 1.6), our study revealed a lower risk ratio (1.38) 12,13 . Other than ethnic differences, the lower risk may come from our study design with controlled confounding factors in the hazard ratio analysis. The systemic inflammation of RA has a direct effect on the endothelium, and predisposes patients to accelerated atherosclerosis 24,29 . Previous studies have shown that cytokines, chemokines, and autoantibodies oxidized low-density and matrix metalloproteinases (MMPs) are pathogenesis factors of atheroscelrosis [30][31][32][33]43 . In addition, smoking is a common risk factor for RA and stroke. Furthermore, RA can involve the heart valve structure and lead to atrial fibrillation, which is a well-documented risk factor for stroke 34 .
Our study revealed SLE as a risk factor for stroke, which is consistent with another population-based study conducted in Taiwan. SLE patients show a higher prevalence of traditional risk factors for atherosclerosis (i.e. hypertension, diabetes mellitus, and smoking) than the general population 16 . The associated anti-phospholipid syndrome can increase the risk of stroke 35,36 . In addition to the systemic inflammation process, lupus patients have been proven to have increased blood viscosity 37 , elevated homocysteinlevels 38 , and susceptible genetic background 39,40 . These factors have been shown to be associated with increased stroke risk. Furthermore, immunosupressants may be a factor for stroke. Leflunomide and cyclosporine may complicate hypertension 41 . However, these studies were only correlation studies, and did not prove causality. Further cohort studies are necessary to verify the effects of these factors. Therefore, we analyzed the effect of steroid usage in SLE patients for developing ischemic stroke. Our study found that when steroid was enrolled as a confounding factor of adjusted HR analysis, SLE was not a risk factor for ischemic stroke. This indicated that the steroid usage can lead the SLE patients more vulnerable to ischemic stroke. Doria et al found a correlation between cumulative steroid dosage and subclinical carotid atherosclerosis in lupus patients 42 . And the outcome of our Indicator *P , .05, **P , .01 and ***P , .001compared with the controls.
www.nature.com/scientificreports study is compatible with previous study by Doria et al. SLE patients with steroid usage should be cautioned with ischemic stroke incidence and efforts for stroke prevention is recommended. For better accuracy of ICD-9-CM diagnostic codes for stroke, we applied the stroke guideline for standard diagnosis and treatment. In Taiwan the Guideline for stroke has been setup for clinical application and Taiwan Stroke registry has been promoted by Taiwan Stroke Association since 2003. Besides, the accuracy of ICD-9 CM coding was checked by committees of the Bureau of the NHI that randomly sample the claim data and review the chart to verify the diagnostic validity regularly. The NHI claim database is a well established research database and has been used in various biomedical research fields. And the accuracy of the NHIRD (National Health Insurance Registration Database) in recording ischemic stroke diagnoses was high (97.85%), and the NHIRD appears to be a valid resource for population research in ischemic stroke. For the efforts of accurate register data acquirement, diagnosis of RA, and SLE were established by specialists in rheumatology based on clinical manifestations and laboratory data, which correlated with the American College of Rheumatology (ACR) 1997 revised criteria. In Taiwan, patients with these rheumatic diseases can apply for catastrophic illness registration cards from the Bureau of NHI and they do not need to additional copayment of using medical resources. For prevention of addition burden of medical insurance, the ICD9-CM coding for these rheumatic diseases is checked by the committees of the Bureau of NHI and the catastrophic illness registration card is allowed after the step of confirmation. In additional to confirmation by rheumatologist and the Bureau of NHI, we only enrolled consecutive coding cases for prevention of wrong coding in the database. These methods could improve the accuracy of these rheumatic diseases registration.
Nevertheless, this study is subject to several possible limitations. First, the diagnosis of SLE, RA, stroke, and medical comorbidities was entirely determined using the ICD codes from the National Health Insurance claim database and there may be concern about the diagnostic accuracy of the database. Even the accuracy of the National Health Insurance Registration Database (NHIRD) in recording ischemic stroke diagnoses was high (97.85%), and it appears to be a valid resource for population research in ischemic stroke 43 . However, there is limited information about the accuracy of RA, and SLE in NHIRD. Besides delayed coding could be occurred of the NHIRD and disease duration before this database is uncertain. These could influence the developing of strokes. Second, the NHIRD does not contain clinical data on the neurologic functional status of patients, or the location and size of cerebral infarct and hemorrhage. And the stratified of the rheumatic disease severity is limited in the NHIRD because the laboratory data and clinical scale cannot be represented. Third, the NHIRD lacks data on individual behaviors that are major risk factors for stroke such as smoking or alcohol consumption. It also does not contain the death records of enrollees. Absence of death records, however, could affect our study results if some of the deaths were stroke-related. Finally, to make precise diagnoses, we adopted various sampling methods in this study. This may result in the prevalence of these 2 rheumatic diseases not being represented collaboratively in this study.

Conclusion
RA is risk factor for ischemic stroke in this population-based longitudinal cohort. And steroid is a crucial factor for contributing the risk for patients with RA and SLE and comparison cohort after adjustment for age, sex, diabetes mellitus, hypertension, coronary heart diseases, hyperlipidemia, congestive heart failure, hypercoagulability, renal disease, atrial fibrillation, valvular heart disease, Aspirin use, Hydroxychloroquine use and NSAIDs use.