This was a retrospective study of a prospective cohort of ischemic stroke patients designed to evaluate the usefulness of MDCT. This cohort included consecutive patients with acute cerebral infarction or transient ischemic attack within 7 days after symptom onset who were admitted to the neurology department of the Severance Stroke Center, Yonsei University in Seoul, Korea. Upon admission, all patients were thoroughly evaluated to determine demographic data, medical history, clinical manifestations, and vascular risk factors.14 All patients underwent brain CT and/or magnetic resonance imaging with cerebral angiographic studies, standard blood tests, and 12-lead electrocardiography. Most patients were admitted to a stroke unit and were monitored continuously with electrocardiography during their stay (average 4.9 days). Holter monitoring was also performed if a cardiac embolism was suspected on the basis of infarction pattern, patient age, or previous cardiac history. Transesophageal echocardiography was a part of the routine examination, unless it could not be performed due to either the patient’s condition or failure to obtain informed consent.

For determination of the presence of asymptomatic CAOD in stroke patients, since July 2006, MDCT was consecutively performed when a patient had at least one of the following: 1) presence of atherosclerosis in an intracranial or extracranial cerebral artery; 2) presence of ≥2 risk factors for CAOD, such as hypertension, diabetes mellitus, dyslipidemia, cigarette smoking, and central obesity; and 3) old age (males >45 years, females >55 years).12 MDCT was not performed if patients had 1) known CAOD; 2) high pulse rates that were not controlled with beta-blockers at the time of MDCT; 3) poor cooperation; 4) impaired renal function; or 5) failure to obtain informed consent. When significant (≥50%) stenosis of the coronary artery was observed on MDCT, the patients were routinely referred to cardiologists at our cardiovascular center. CAOD was then managed at the discretion of the cardiologists and neurologists, which included percutaneous coronary intervention with stent (PCI) and coronary artery bypass graft (CABG). This study was approved by the Institutional Review Board of Severance Hospital, Yonsei University Health System. Informed consent was waived due to the retrospective nature of the study.

We collected data on vascular risk factors for coronary heart disease, such as hypertension, diabetes mellitus, hyperlipidemia, and current smoking. The presence of valvular heart disease, atrial fibrillation, congestive heart failure, previous ischemic or hemorrhagic stroke, peripheral arterial occlusive diseases, metabolic syndrome, underlying malignancy (cancer diagnosed within 6 months prior to the index stroke, ongoing treatment for cancer, or recurrent or metastatic cancer), chronic kidney disease (estimated glomerular filtration rate < 30 mL/min/1.73 m²), and prior medication before the index stroke were also investigated. Initial stroke severity was determined based on the National Institute of Health Stroke Scale (NIHSS). Cerebral atherosclerosis was defined as having at least ≥1 steno-occlusive lesion of >50% based on the North American Symptomatic Carotid Endarterectomy Trial (for extracranial arteries)15 or on the Warfarin vs. Aspirin for Symptomatic Intracranial Disease method (for intracranial arteries).16 Framingham risk scores were calculated for each patient. Data on lipid profiles were also obtained.

Outcomes measured in this study were death and first occurrence of cardiovascular events or recurrent ischemic stroke. We obtained data for vascular events and death (date and causes of death) by review of the medical records and face-to-face or telephone interviews carried out by the stroke research nurse and stroke specialists. For mortality data, the date and cause of death were also identified using data from the Korean National Statistical Office. These mortality data are known to be reliable because they are collected based on a unique 13-digit identification code assigned to subjects at birth and on the death certificate. The causes of death are coded according to the International Classification of Disease, 10th Revision (ICD-10). Cardiovascular events were defined as nonfatal or fatal myocardial infarction (ICD-10 code I20-21), chronic ischemic heart disease (I25), sudden cardiac death (R96), death due to ventricular fibrillation (I49), or congestive heart failure (I50). The ICD-10 code used for coding fatal or nonfatal ischemic stroke was I63. The date of censoring was December 31, 2012.

Continuous variables are expressed as mean±standard deviation, and categorical data are expressed as number (percentage). The chi-square test or Student’s t-test was used to compare groups for categorical variables and continuous data, as appropriate. A Kaplan-Meier estimate of survival was used to compare the differences in rates of cardiovascular events or recurrent ischemic stroke between the MDCT (+) group and the MDCT (−) group. Differences were determined by the log-rank test. To identify independent factors for the events, multiple Cox regression analysis was used.

Further, to reduce the potential effects of selection bias and confounding factors in this retrospective cohort study, we estimated propensity scores for each of the entire patients enrolled in this study to match the patients who had undergone MDCT to those who had not. This was computed for each patient with a logistic regression model, including variables of age, sex, hypertension, diabetes, hypercholesterolemia, current smoking status, valvular heart diseases, atrial fibrillation, congestive heart failure, previous ischemic or hemorrhagic stroke, peripheral arterial occlusive diseases, metabolic syndrome, underlying malignancy, chronic kidney disease, prior antiplatelet/anticoagulant/statin use, initial stroke severity, concomitant cerebral atherosclerosis, and lipid profiles. Each patient was then assigned an estimated propensity score, which was the predicted probability of undergoing MDCT on the basis of patient’s observed baseline characteristics. Then, we divided the cohort into five strata defined by quintiles of estimated propensity scores. Cox proportional hazards models were employed separately within each stratum to compare the overall survival of patients undergoing MDCT with that of patients not undergoing MDCT. The five hazard ratios (HR) estimated from each stratum were combined into an overall HR for the whole cohort.

We also used Cox models for adjusting differences between the groups in other ways. First, regression adjustment was performed with inclusion of the propensity score as a linear predictor in the model. Second, the inverse probability of treatment weighting was used. We used Cox regression analysis to adjust the inverse probability of treatment weighting, as well as baseline characteristics. Finally, propensity score matching was conducted between paired patients who underwent MDCT and those who did not undergo MDCT. After matching, the baseline characteristics were compared with McNemar’s test and the paired t-test. Clustered Cox regression analysis was also performed. All calculated p-values were two-sided, and p<0.05 was considered to be statistically significant. Statistical analyses were performed using the Windows IBM SPSS software package (version 20, IBM Corp., Armonk, NY, USA) and SAS (version 9.2, SAS Inc., Cary, NC, USA).

A total of 4381 consecutive patients were registered between July 2006 and December 2012. This comparative analysis was designed to investigate the usefulness of MDCT among patients who had no previous history of CAOD. Consequently, of the 4381 patients, we first excluded 938 patients who had a history of CAOD (myocardial infarction, angiographically confirmed CAOD, unstable angina, PCI, or CABG) or who underwent coronary angiographic evaluation within 1 year prior to the index stroke. We further excluded 160 patients who did not have any of the inclusion criteria for screening of asymptomatic CAOD with MDCT. Patients with unavailable data on initial NIHSS scores (n=19), estimated glomerular filtration rate (n=20), or incomplete lipid profiles (n=100) and those who died of acute index stroke within 7 days of symptom onset (n=27) were also excluded. Finally, 3117 patients were included in this study. Among them, 1842 patients underwent MDCT and 1275 patients did not undergo MDCT (Fig. 1). In the MDCT (−) group (n=1275), the reasons for not performing MDCT could be determined as follows: medical reasons (52.5%, 669/1275), such as poor cooperation or general condition (34.0%), renal or contrast agent-related factors (10.0%), and rapid heart rate (8.5%); and nonmedical reasons (47.5%, 606/1275), including a failure to obtain informed consent or no specific documented reasons.

The mean age of the 3117 ischemic stroke patients enrolled in this study was 65.7±12.0 years, and 60.5% of patients were men (n=1886). On the basis of Framingham risk scores, 28.6% of patients were classified into a low risk group (10-year risk <10%), 33.2% into a moderate risk group (10–20%), and 38.2% into a high risk group (>20%). In the MDCT (+) group (1842 patients), asymptomatic CAOD was detected in 1237 patients (67.2%): minimal CAOD in 725 (39.4%), 1-vessel disease in 301 (16.3%), 2-vessel disease in 149 (8.1%), and 3-vessel disease in 62 (3.4%).

Compared to patients of the MDCT (+) group, those of the MDCT (−) group were more likely to be older, female, and non-smokers and to have hypertension, valvular heart disease, atrial fibrillation, congestive heart failure, previous ischemic stroke, peripheral arterial occlusive diseases, underlying malignancy, chronic kidney disease, prior anticoagulant treatment, more severe stroke, concomitant cerebral atherosclerosis, and relatively low triglyceride, low density lipoprotein, or high density lipoprotein levels (Table 1). In terms of medication at discharge, statins were more frequently prescribed in the MDCT (+) group (n=1773, 96.3%) than in the MDCT (−) group (n=1148, 90.0%) (p<0.001). There were similar trends in prescribing antithrombotics [n=1810 (98.3%) in MDCT (+) vs. n=1202 (94.3%) in MDCT (−), p<0.001] or beta-blockers [n=631 (34.3%) in MDCT (+) vs. n=287 (22.5%) in MDCT (−), p<0.001].