Infammatory bowel diseases increase future ischemic stroke
risk: A
Taiwanese population-based retrospective cohort study
Wei-Shih Huang
a,b,1, Chun-Hung Tseng
a,b,1, Pei-Chun Chen
c,d, Chon-Haw Tsai
a,b, Cheng-Li Lin
e,f,
Fung-Chang Sung
e,f,⁎
, Chia-Hung Kao
b,g,⁎⁎
1. Introduction
Stroke results in the sudden loss of neurological function caused by a disruption of blood fow to the brain. Previous studies have reported
that stroke is the second leading cause of death worldwide [1–3]. In
Taiwan, ischemic stroke is themajor type of stroke [3,4],which is similar
to reports from the rest of the world. Stroke is also among the leading
causes of disability in adults [5]. The identifcation of risk factors for
stroke is crucial for the primary and secondary prevention of stroke. The crucial role of infammatory mechanisms in the progression of atherosclerosis has promoted many studies focusing on whether diseases characterized by chronic infammation, including infammatory bowel disease (IBD), carry an increased risk of cardiovascular
disease [6,7]. Numerous previous studies have reported an increased
incidence of stroke and myocardial infarction (MI) in patients with rheumatoid arthritis, psoriasis, and systemic lupus erythematosus
[8–10], but studies on the risk of atherothrombotic disease among
patients with IBD are inconclusive [11–15]. However, recent evidence
showed that IBD is associated with an increased risk of stroke and MI
[16,17].
IBDmainly comprises the following two chronic intestinal disorders: Crohn's disease (CD) and ulcerative colitis (UC) in addition to microscopic colitis. IBD affects approximately 1.4 million people in the
United States [18] and approximately 2.2 million people in Europe [19]. Chuang et al. recently reported
that IBD is still relatively uncommon
in Taiwan, but the incidence and prevalence rates are increasing
[20]. The linkage between IBD and atherothrombotic disease could
have a substantial infuence on patientmanagement. Therefore,we conducted this nationwide population-based retrospective cohort study
to assess the association between IBD and the risk of future ischemic stroke among adult patients in Taiwan.We hypothesize that IBD is associated
with the development of acute ischemic stroke and collected the study data from Taiwan's National Health Insurance Research Database (NHIRD) among adult patients (≥20 years old).
2. Patients and methods 2.1. Data sources
The National Health Insurance program has been implemented in Taiwan since 1995 and provides medical cover for approximately 99%
of Taiwan's population (23.74 million people) [1]. The National Health
Research Institutes (NHRI) of Taiwan established the National Health Insurance Research Database (NHIRD), which contains registration fles and original claims data for reimbursement. The NHRI manages the annual claims data in the NHIRD and has established electronic datasets for administrative and research purposes. This study used a subset of NHIRD containing one million subjects randomly selected from among population insured by the National Health Insurance program of Taiwan. The diagnoses codes used in the NHRI were based on
the International Classifcation of Diseases, 9th Revision, ClinicalModifcation (ICD-9-CM). The NHRI encrypts the patients' personal data to
protect their privacy. This study was approved by the Ethics Review Board of China Medical University (CMU-REC-101-012).
2.2. Study patients
The IBD cohort comprised extracted data for newly diagnosed IBD patients with either CD (ICD-9 codes 555.0–555.2) or UC (ICD-9 code 556) from 1998 to 2011 who had received either ambulatory or inpatient care.We used the date of IBD diagnosis as the index date.We randomly selected a comparison cohort comprising insured peoplewithout
a history of IBD. For each corresponding IBD patient, we selected 4 control patients matched for age, sex, and year of index date.We excluded patients diagnosedwith stroke (ICD-9 code 430–438) prior to the index date, those who were younger than 20 years, or cases lacking age and sex data and the ischemic stroke cases diagnosed within one year after the index date.
2.3. Outcome defnition and comorbidities
To identify the newly diagnosed cases of ischemic stroke (ICD-9 code 433–438), we used the patient identifcation number to link the study patients to their corresponding inpatient claims. We calculated the follow-up person-years for each study patient from the index date until that patient was diagnosed with ischemic stroke, lost to follow-up,
died, or withdrew from the insurance system, or until the end of 2011. Preexisting comorbidities for each patient included hypertension (ICD-9 codes 401–405), diabetes (ICD-9 code 250), coronary artery disease (CAD) (ICD-9 codes 410–414), congestive heart failure (CHF) (ICD-9
codes 398.91, 402, 404.01, 404.03, 404.10, 404.11, 404.13, 404.9), atrial fbrillation (ICD-9 codes 427.31, 427.32), and peripheral arterial disease
(PAD) (ICD-9 codes 440.2, 440.3, 444.2, 444.81). We also estimated whether the patients had any of these comorbidities.
2.4. Statistical analyses
We compared specifc demographic factors, including age, sex, and comorbidities, between the IBD and non-IBD cohorts by using the chisquare test for the categorical variables and the student's t test for the
continuous variables. We plotted the ischemic stroke-free survival curves by applying the Kaplan–Meiermethod, and the level of statistical signifcance was based on the log-rank test. For each cohort, we calculated the sex- and age-specifc incidence rates of ischemic stroke
per 1000 person-years of follow-up. We used the Poisson regression model to estimate the incidence rate ratios (IRRs) for the IBD and
non-IBD cohorts by using 95% confdence interval (CIs). To assess the effects of IBD on the risk of ischemic stroke,we used themultivariate Cox
proportional-hazards regression model, and adjusted for any variables that related signifcantly to ischemic stroke. We included IBD-related medical visits, defned by outpatient visits and/or hospital admissions with ICD diagnoses of IBD recorded in primary or secondary position on the claims. For each IBD subject, the number of annualmedical visits was calculated by using number of the IBD-related medical visits during the entire follow-up period divided by number of the follow-up personyears. Patients were then classifed into 3 categories, 1–2, 3–4 and N4
times, according to their mean number of IBD related visits per year. We used the Cox model to calculate the hazard ratios (HRs) and 95% CIs. To estimate the IRR and HR of ischemic stroke associated with the frequency ofmedical visits for IBD,we used both the Poisson regression model and Coxmodel to compare the IRR and HRwith those of the non-IBD cohort.We used SAS (version 9.2 forWindows; SAS Institute, Cary, NC, USA) to conduct all statistical analyses. The Kaplan–Meier survival curve was plotted using R software (version 2.14.1; R Development Core Team, Vienna, Austria). P values b .05were considered statistically signifcant.
3. Results
The IBD and non-IBD cohorts comprised 18 392 and 7 3568 cases, respectively.
Among the study patients, 64.5% were younger than 50 years,
and 54.9% were women (Table 1). The mean ages of the IBD and
non-IBD cohorts were 44.8 ± 16.6 and 44.5 ± 16.8 years, respectively.
Compared with the non-IBD cohort, the IBD cohort had a higher prevalence of hypertension (22.7% versus 19.0%, P b 0.001), diabetes (9.73%
versus 8.11%, P b 0.001), CAD (12.2% versus 8.66%, P b 0.001), and cCHF (9.67% versus 7.80%, P b 0.001). The results of the Kaplan– Meier analysis showed that the overall ischemic stroke-free rate was signifcantly lower among the IBD cohort than among the non-IBD cohort
(log-rank P = 0.003; Fig. 1) throughout the entire 14-year study
period. The overall incidence rate of ischemic stroke was 1.15-fold higher among the IBD cohort than among the non-IBD cohort (4.31 versus
4.35 per 1000 person-years), and the adjusted HRwas 1.12 (95% CI, 1.02–1.23) (Table 2). The
sex-specifc analysis results show that the
incidence rate was higher among men than among women for both cohorts and age-specifc analysis results show that the incidence rate increased in conjunction with age for both cohorts. Moreover, these results show a signifcantly higher risk of developing ischemic stroke among older adults (ie, 50–65 y) (adjusted HR, 1.23; 95% CI,
1.04–1.46). Table 3 shows the incidence and adjusted HR of ischemic
stroke associated with different entities of IBD. Compared with the
non-IBD cohort, the CD patients exhibited a relatively higher risk of developing
ischemic stroke (adjusted HR, 1.15; 95% CI, 1.04–1.28). Table 4
shows the association between the frequency of medical visits for IBD
patients and the development of ischemic stroke. The overall risk of developing ischemic stroke increased from 1.06 to 3.87, which correlates
positively with the frequency of visits (trend test P b 0.0001). Compared to the non-IBD cohort, the adjusted HR increased from 1.07
(95% CI, 0.97–1.18; ≤2 CD-related visits) to 6.36 (95% CI, 3.83–10.6; N4 CD-related visits) (trend test P b .0001); and the adjusted HR also increased from 1.11 (95% CI, 1.01–1.22; ≤2 UC-related visits) to 2.10
(95% CI, 1.05–4.20; N4 UC-related visits) (trend test P = .01). We assessed the HR stratifed by median follow-up period. The follow-up period was partitioned by median (within 7 years, and more than
with the follow-up periods between 2 cohorts. We found a
1.15-fold signifcantly higher relative risk of developing ischemic stroke within 7 years follow-up period (95% CI= 1.03–1.28).
4. Discussion
This study shows that the IBD cohort was at a 1.12-fold (95% CI,
1.02–1.23, Table 2) higher risk of developing subsequent ischemic
stroke than the comparison cohort after adjusting for sex, age, and comorbidities. Comorbidities such as hypertension, diabetes, CAD, CHF,
peripheral arterial disease, atrial fbrillation, and hyperlipidemia were signifcant factors associated with the occurrence of ischemic stroke. Compared with the UC patients in this study, the CD patients had a higher risk of developing ischemic stroke (the adjusted HR of CD
versus UC is 1.15:1.01) (Table 3). We further observed a strong association
between increased number of annual IBD-related medical visits and the risk of developing ischemic stroke. IBD patients with more frequently visits to hospitals may represent more IBD fare-up or more duration and severity of the active periods. These results indicate that the systemic infammatory burden among the IBD patients may be a crucial determinant of atherothrombotic risk. These fndings
also support previous studies that have shown a disease severitydependent increase in risk of developing stroke and MI among patients
with other chronic infammatory diseases such as rheumatoid arthritis
and psoriasis [21,22]. Surveillance bias might occur, as patients with
IBD who seek medical care more frequently may be more likely to be aware of adverse health conditions. However, a previous report has shown that the prevalence of silent stroke was only 10% in middleaged population with a mean age of 62 years, and the prevalence
was lower in younger adults [34]. Our analysis including population
with a mean age of 45 years might have a prevalence of silent stroke lower than 10%. Furthermore, the association between number of
IBD-relatedmedical visits and stroke was strong, with 3.87-fold risk associated with N4 IBD visits and 6.36-fold risk associated with N4
Crohn's disease. The observed association is not likely to be fully explained by the surveillance bias.
The peak onset of IBD is usually in persons 15–30 years of age [18].
An increasing amount of evidence has indicated that IBD results from an unusual infammatory response to intestinalmicrobes in a genetically
it can affect any region of the intestine, often discontinuously.
UC involves the rectum and may affect either part of or the entire colon (pancolitis) in an uninterrupted pattern. The increased risk of
venous thromboembolic events in patients with IBD is now wellestablished
[27], but the risk of atherothrombotic diseases such as stroke
andMI have been the topic of debate [11–15]. For example, a 2007metaanalysis
of 11 studies comprising approximately 14,000 IBD patients
showed no increased risk of cardiovascular mortality [13]. However, a
cohort of 8054 CD patients from the UK General Practice Research Database exhibited a nearly 3-fold (odds ratio, 2.93 [1.44–5.98]) increased
risk of developing stroke among patients younger than 50 years of age,
but not in the elderly [11]. Singh et al. recently also reported that IBD is
associated with a modest increase in the risk of stroke and ischemic
heart disease in a meta-analysis [17].
Our results contribute to extant literature by showing a signifcantly increased risk of ischemic stroke among a relatively large and unselected population of IBD patients in Taiwan. A 2013 nationwide population study in Denmark also showed a signifcantly increased risk of developing MI, stroke, and cardiovascular death among a
large and unselected population of IBD patients [16]. A novel fnding
from that study was that the risk was related to IBD activity, and the highest risk was exhibited during fares and periods of persistent activity, whereas the risk of MI and stroke was increased only marginally during remission periods; in the latter periods the risk of cardiovascular death in IBD patients was comparable to that of the
reference population. Those fndings agree with our results that show a parallel signifcant trend (P b .0001) between the relative increase in the risk of developing ischemic stroke and the frequency of medical clinic visits, regardless of the IBD entity; moreover, the increased frequency of medical clinic visits in our study could indicate the severity
of the disease.
Atherosclerosis is a chronic infammatory disease characterized by both infammation of the arterial wall and systemic infammation of the body. IBD is also characterized by an unusual immuno-infammatory
activation. Furthermore, there are similarities between the pathophysiological processes in the colonic wall of IBD patients, the processes
in the arterial wall during the progression of atherosclerosis and, ultimately, atherosclerotic plaque rupture and thrombotic vascular events
[26,28,29]. Previous reports that have associated IBD with subclinical atherosclerosis, including endothelial dysfunction and increased carotid intima-media thickness, coupledwith the atherogenic alterations of the lipid profle, lend additional support to our fnding of an increased risk
of developing ischemic stroke among IBD patients [30–33].
However, our study has limitations. First, the NHIRD does not
provide data on detailed patient information such as smoking habits, alcohol consumption, body mass index, physical activity, socioeconomic
status, family history, andmedication use,which are all potentially confounding factors for this study. Second, unlike randomized trials, a retrospective cohort study design is subject to biases related to confounder
adjustments. Although our study design adequately controls for confounding factors, the potential bias caused by unmeasured or unknown
confounders could be a key limitation. Third, our analysis included diagnosed IBD patients, as we relied on claims data that do not cover subjects with mild symptoms who did not seek care.
In summary, compared to the non-IBD patients, our study shows the IBD patients are at an increased risk of developing ischemic stroke, which increased in conjunction with the frequency of medical visits for both UC and CD groups.. This fnding provides evidence that effective treatment of IBD aimed at disease activity remissionmay reduce the risk of developing ischemic stroke among patients with IBD.
