A nationwide population-based retrospective
cohort study: increased risk
of acute coronary syndrome in patients with
ankylosing spondylitis
C-H Chou1*, M-C Lin2*, C-L Peng3,4, Y-C Wu2, F-C Sung3,4, C-H Kao5,6, S-H Liu7,8
Acute coronary syndrome (ACS) includes conditions causing sudden reduced blood flow to the heart. ACS refers to unstable angina, non-ST-segment elevation myocardial infarction, and ST-segment elevation myocardial infarction (1, 2). The causes of cardiovascular disease (CVD) are multifactorial (3, 4); some of the CV risk factors are male gender, hypertension, smoking, hyperlipidaemia, and diabetes mellitis (DM). Ankylosing spondylitis (AS) is a
chronic and systemic inflammatory condition that primarily affects the sacroiliac joints and the axial skeleton (5–8). Increased risk of CV-related morbidity and mortality has been reported in patients with inflammatory arthritis (9, 10). The inflammatory processes in rheumatoid arthritis (RA) are similar to atherosclerosis, including the formation of atheroma, plaque instability, and thrombus development (11). One study showed that an elevated C-reactive protein (CRP) concentration measured early in the disease process is a powerful predictor of death from CVD in inflammatory polyarthritis and RA (12). Data from USA have shown an increased prevalence of CV disease in AS patients compared with age- and gender-matched controls (13). Another
study found a higher prevalence of spondyloarthropathies among patients receiving a coronary artery bypass graft (CABG) compared with control CABG patients, and
demonstrated that spondyloarthropathy was a stronger risk factor for CABG than most traditional risk factors (14). Despite these studies, published reports of the relationships between AS and CVrisk factors are limited (15). Hence, the
aim of our study was to compare the incidence of ACS and CV risk factors between patients with and without AS. Method
Data sources
Our retrospective cohort study used reimbursement data
from National Health Insurance (NHI) electronic records system in Taiwan, which contains all medical claims from
1996 to 2009. In Taiwan, 99% of citizens are covered under the NHI and 90% of hospitals and clinics are NHI contracted providers. The National Health Research
Institutes (NHRI) is responsible for managing the insurance claims data reported to the Bureau of Health
Insurance. The NHRI has established electronic datasets for administrative and research purposes. We used a subset of data that comprises claims data from a random
sample from the one million insured people in the nationwide database compiled by the NHRI and released for us
to use in 2009. The diagnosis codes used in the NHRI are based on the International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM). Patients’ personal data are encrypted in this dataset for protection of privacy. This study was approved by the Ethics Review Board of the China Medical University and Hospital (CMU-REC-101-012).
Study subjects and outcome measures
Our study consisted of an AS cohort and a non-AS cohort identified from the claims data of 2000–2009. Both cohorts were followed up until the end of 2009. The AS
cohort consisted of patients aged _ 18 years newly diagnosed with AS in 2000–2009, identified from both ambulatory
care and in-patient care units (ICD-9-CM codes
720 and 7200). The index date for an AS patient was the date of the first clinic visit with the diagnosis. Subjects with a history of ACS diagnosed (ICD-9-CM 410–411.1) before the index date or with missing information on age or sex were excluded. The comparison cohort was
selected using a systematic random-sampling method in which four control patients, matched for age, sex, and
year of index date, were selected for each corresponding AS patient. The follow-up person-years were calculated for each study participant from the index date to the date of withdrawal from the insurance system or to the end of 2009. All study subjects with a diagnosis of ACS or CVD risk factors were already confirmed by ICD-9 coding (ICD-9-CM 410–411.1) according to the study subject’s medical records in the National Health Insurance
Research Database (NHIRD), based on based on validated definitions (16).
Statistical analysis
The distributions of categorical sociodemographic characteristics were compared between the AS cohort and the
non-AS cohort, and the differences were examined using the χ2 test. The follow-up person-years were used for
estimating the incidence density and the AS cohort to non-AS cohort incidence rate ratio (IRR) calculations used Poisson regression. Cox’s proportional hazards regressions were used to assess the risk of developing ACS that was associated with AS, while adjusting for variables that were significantly related to AS based on the results of the χ2 test. Hazard ratios (HRs) and 95%
confidence intervals (CIs) were estimated in the model. All analyses were performed SAS version 9.2 (SAS Institute, Cary, NC, USA). A two-tailed p value < 0.05 was considered to be statistically significant.
Results
We selected 6262 patients with AS for our study cohort and 25 048 control patients without AS for our comparison cohort. Among the study participants, 52.3% were
women and 83.1% were younger than 65 years of age. The most common co-morbidities in the AS cohort compared with the non-AS cohort were hypertension (26.4%
vs. 20.0%, p < 0.0001), DM (8.7% vs. 7.3%, p < 0.0001), hyperlipidaemia (23.3% vs. 15.8%, p < 0.0001), and stroke (14.3% vs. 10.0%, p < 0.0001) (Table 1). Table 2 shows the incidence rates of ACS for both cohorts, the AS cohort to the non-AS cohort IRRs and
HRs by sex and age. The incidence rate of ACS was 1.5-fold higher in the AS cohort than in the non-AS cohort (4.4 vs. 2.9 per 1000 person-years) with an adjusted HR
(aHR) of 1.36 (95% CI 1.16–1.59). The sex-specific AS cohort to non-AS cohort IRRs were similar for both men
and women, as were the HRs. The incidence of ACS increased with age in both cohorts, being greater in those with AS and the highest in the oldest group. However, the age-specific AS cohort to non-AS cohort IRR and aHR of ACS were the greatest for the younger patients (aged 18–44 years). The age-specific aHR was not significant for the older groups.
Figure 1 shows the Kaplan–Meier graph of ACS-free survival rates for AS patients and the comparison cohort. The results of the log-rank test indicate that patients with AS had a significantly lower ACS-free survival rate
(95.8% vs. 97.1%, p < 0.001).
Table 3 shows that subjects with co-morbidity had a higher incidence of ACS than those without co-morbidity, which was consistently higher in the AS cohort than in the non-AS cohort. Table 4 shows the interaction measures between AS and selected co-morbidities for the risk of ACS. Hypertension was most prevalent in both AS and non-AS cohorts, and was associated with an AS to non-AS aHR of 2.84. The corresponding aHR was 2.58 for those with DM. The aHR increased to 4.36 for those with both hypertension and DM. Cancer was the least prevalent in AS patients. However, AS patients with hypertension, DM, and cancer had an aHR of 7.74.
Discussion
In our study, the risk of ACS was higher in the AS cohort than the non-AS cohort. We also found a significantly higher prevalence of hypertension, DM, hyperlipidaemia, and stroke in patients with AS than in control patients. A previous meta-analysis showed that the risk of myocardial infarction in AS patients may be related to an
increased frequency of dyslipidaemia (17). In a recent retrospective cohort study, although an increased risk of
myocardial infarction in AS patients was not observed, the prevalence of DM and hypertension was higher for AS patients than for the controls (18).
Health-care databases have been used in CV studies to assess disease prevalence and costs, and the reliability of these data resources has been deemed adequate to track the quality of health care (19–22). The use of ICD-9 codes to identify CVDs and their risk factors have demonstrated positive predictive values for most CV events, especially for acute myocardial infarction and stroke (23–25). In our analysis, a higher risk of ACS was shown in AS patients, compared with controls. Similarly, Peters et al (26)
reported that the overall prevalence for myocardial infarction was 4.4% in patients with AS and 1.2% in the general
population, resulting in an age- and sex-adjusted odds ratio of 3.1 (95% CI 1.9–5.1) for patients with AS. Previous studies have shown that AS is diagnosed
more frequently in males than females, with a male-tofemale ratio of 3:1 (27, 28). Another study showed a
male-to-female ratio of 1:3 for undifferentiated spondyloarthropathy (29). The male-to-female ratio of our study
was approximate 1:1. Previous studies have reported that females may have milder symptoms or manifest subclinical disease more frequently than males. Thus, differences
in health-care-seeking behaviors between our sample and those of previous studies may have contributed to the
diversity of results.
It is a reasonable that older patients would have a higher risk of ACS in the general population. In our study, the oldest people (age _ 75 years) had the highest incident rates of ACS and these were higher in the AS cohort than in the non-AS cohort (17.3 vs. 13.3 per 1000 person-years; Table 2). In addition, it is known that younger people have a lower incidence of hypertension, DM, and hyperlipidaemia. However, the age-specific AS to non-AS IRRs and aHRs of ACS were the largest in the youngest group. Therefore, younger people are more likely to be affected by AS only.
The strengths of the present study included its use of population-based data, which cover a highly representative sample of Taiwan’s general population under a universal reimbursement policy. However, certain limitations should be mentioned. First, some important data are missing from the NHIRD, including detailed demographic and lifestyle information such as smoking habits, alcohol consumption, body mass index, socio-economic status, and family history of systemic diseases. These are major risk factors for ACS. Second, the evidence derived from a retrospective cohort study is generally lower in quality than that from randomized trials because the former is subject to many biases related to adjustments for confounding variables. Despite our meticulous study design with adequate control of confounding factors, a key limitation is that bias could remain if unmeasured or unknown confounders were present. Third, because all beneficiaries listed on the NHIRD
are protected by anonymity, we could not obtain individual clinical (blood pressure), imaging, pathological, and laboratory data [including high density lipoprotein (HDL)
cholesterol, fasting glucose, triglycerides: metabolic syndrome; inflammatory markers: CRP or erythrocyte sedimentation
rate (ESR); and disease activity: Bath Ankylosing Spondylitis Disease Activity Index (BASDAI)] for each
study subject. Fourth, we were unable to contact the patients
directly to enquire about their use of non-steroidal antiinflammatory drugs (NSAIDs), details of which were not
available from the database. Therefore, we could not evaluate whether medications for AS patients altered their risk of ACS. Further larger population-based studies or large-scale randomized clinical trials to confirm our findings are necessary before any definite conclusions can be drawn.
In conclusion,AS patientsmay be at higher risk for ACS comparedwith the general population.As recently reported in the European League Against Rheumatism (EULAR) evidence-based recommendations (30), management of CV risk factors and control of systemic inflammation should be taken into account in patients with AS.
