This study was a nationwide cohort study using the National Health Claims database, established by the National Health Insurance Service (NHIS) of Korea. The NHIS database represents the entire population of South Korea,11 and all citizens are consecutively enrolled unless they are ineligible due to emigration or death. This database contains comprehensive information on health care utilization in Korea, including diagnoses, prescriptions, and surgical procedures. Therefore, it provides detailed data on personal characteristics and health care utilization based on reimbursed claims for inpatient, outpatient, and emergency department visits. This study was approved by the Institutional Review Board of Samsung Medical Center. Informed consent was waived as deidentified data were used.

From 2013 to 2020, the use of FFR was reimbursed by NHIS if the patients met the following criteria: a) 50%-70% intermediate stenosis on quantitative coronary angiography in a vessel greater than 2.5mm in diameter and b) no prior evidence of inducible myocardial ischemia on noninvasive tests. The current study included participants in whom FFR use was reimbursed based on these criteria, but whose revascularization was deferred between 2013 and 2020 (n=12 026). To include only patients with intermediate coronary artery stenosis, we excluded cases reimbursed after the 2021 expansion of the criteria from 50%-70% to 50%-90% stenosis. Exclusion criteria consisted of patients with a prior history of any bleeding requiring blood transfusion (n=193), atherosclerotic cardiovascular disease (ASCVD, n=5781), or those already on nonvitamin K antagonist oral anticoagulants (NOAC) or warfarin (n=1070), and patients taking any antiplatelet agents between 30 days and 365 days before the index admission (n=2159). Since participants could meet more than 1 exclusion criterion, we excluded 6683 patients. Among the remaining patients, those who received dual antiplatelet agents or potent P2Y12 inhibitors at discharge (n=686) were also excluded.

Consequently, the final analysis included 4,657 patients. Patients prescribed any antiplatelet agents from discharge and who maintained use for more than 365 days were classified into the antiplatelet therapy group, while the remaining patients comprised the no-therapy group (figure 1).

Figure 1.

Study Flow. This study was a nationwide cohort study of the National Health Claims database established by the National Health Insurance Service (NHIS) of Korea. The NHIS database represents the entire population of South Korea, and all citizens are consecutively enrolled unless they are ineligible due to emigration or death. Among the 4657 eligible patients, 3018 were prescribed antiplatelet agents from the discharge date of the index hospitalization and 1639 were not. After propensity score matching, a total of 3268 patients were included in the current analysis (1634 pairs according to the use of antiplatelet agents). ASCVD, atherosclerotic cardiovascular disease; FFR, fractional flow reserve; NOAC, nonvitamin K antagonist oral anticoagulants; NHIS, National Health Insurance Database.

(0.33MB).

The NHIS database contains data on diagnoses, procedures, length of hospitalization, and prescribed medications, including prescription duration, dosage, route of administration, and costs. All procedures and prescriptions are coded using the Anatomical Therapeutic Chemical Classification system and domestic codes. Diagnoses are coded according to the Korean Standard Classification of Diseases, 7th revision, and a modified version of the International Classification of Diseases, 10th revision (ICD-10). The NHIS routinely audits claims data, which is considered reliable and has been used in multiple previous publications.12,13

Baseline characteristics included age, sex, health behaviors, and comorbidities. Data on income at the time of the first screening examination were obtained from the insurance eligibility database, with income levels categorized as medical aid or not. Residential area at the first screening examination was classified as metropolitan or rural. Health behaviors and laboratory tests were obtained from the national health screening examination database, reflecting the most recent data within 4 years prior to the index hospitalization with FFR measurement. Information on smoking, drinking frequency, and alcohol consumption was also collected via self-administered questionnaires during the national health screening examination. Current alcohol intake (g/d) was calculated based on the frequency of drinking (times per week) and amount of alcohol per occasion, as reported in self-administered questionnairesat the health screening examinations. Heavy alcohol consumption was defined as ≥40g/d in women and ≥60g/d in men.

Comorbidities—including hypertension, diabetes mellitus, dyslipidemia, chronic kidney disease, liver cirrhosis, cancer, and anemia—were identified by diagnosis codes, prescription records, and inpatient and/or outpatient visits within 1 year before the index hospitalization with FFR measurement (table 1 of the supplementary data). Concomitant medications included nonsteroidal anti-inflammatory drugs (NSAIDs), steroids, proton pump inhibitors, and H2 blockers. Long-term use of NSAIDs and steroids was defined as prescription of these medications for at least 90 and 30 days, respectively, between 180 days prior to and 14 days after the index hospitalization. Concomitant use of proton pump inhibitors and H2 blockers was defined as prescription as discharge medication.

The primary efficacy outcome was major adverse cardiac and cerebrovascular events (MACCE), defined as a composite of all-cause death, MI, unplanned revascularization, and stroke. Vital status and cause of death were obtained from death certificates collected by Statistics Korea at the Ministry of Strategy and Finance of South Korea.12 The diagnosis of MI was made if patients were hospitalized with primary diagnostic codes related to MI (ICD-10 I21, I22) during the follow-up period. A previous validation study found that the accuracy of MI diagnosis in NHIS data was 93%.14 Unplanned revascularization was defined as the presence of procedure codes for percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) following the index hospitalization. Stroke was defined based on ICD codes for ischemic stroke (ICD-10 I63, I64) or intracranial hemorrhage (ICD-10 I60-62), in conjunction with codes for diagnostic brain imaging, such as magnetic resonance imaging or perfusion computed tomography. The secondary efficacy outcomes consisted of individual components of MACCE and major adverse cardiac event (MACE), a composite of all-cause death, MI, and unplanned revascularization.

The primary safety outcome was any gastrointestinal bleeding, regardless of the need for transfusion. Secondary safety outcomes were major bleeding, defined as a composite of intracranial bleeding or gastrointestinal bleeding with documented transfusion, gastrointestinal bleeding necessitating hospitalization without documented transfusion, and intracranial hemorrhage. The diagnostic codes in the primary position during hospitalization showed a positive predictive value of 92% for gastrointestinal bleeding and 81.4% for intracranial bleeding, as documented by diagnostic brain imaging.15

Baseline characteristics of the exposed and comparator groups are expressed as frequencies with proportions for categorical variables and as means with standard deviations for continuous variables. The propensity score was estimated for each emulated cohort to minimize systematic differences in baseline characteristics between the 2 groups. All covariates, including age, sex, income, medical aid, residential area, comorbidities, and long-term NSAID or steroid use, were included in the logistic regression model to estimate the probability of receiving treatment, conditional on these covariates. To further minimize bias, we implemented a 1:1 propensity score nearest-neighbor matching with a caliper of 0.2 on the propensity score scale.16 Differences in baseline covariates between the 2 groups were evaluated before and after propensity score matching using an absolute standardized difference, with a value of >0.2 indicating a significant difference.16 In addition, an inverse probability treatment weighting was performed using all eligible participants in the current study as a sensitivity analysis.

An intention-to-treat approach was implemented to investigate the efficacy of the randomly assigned treatment, regardless of treatment adherence. To mimic this approach, patients were followed up from the discharge date of the index hospitalization until the occurrence of a clinical outcome or a prespecified time interval (1 year and 5 years). Within the matched cohort, the 1- and 5-year cumulative incidence of each clinical outcome was estimated using the Kaplan-Meier method, with log-rank tests used to evaluate differences between groups. We calculated hazard ratios (HRs) with 95% confidence intervals (95%CIs) for the incidence of clinical outcomes using the Cox regression model, adjusting for discharge medications, including proton pump inhibitor/H2 blockers. When comparing 3 groups according to the use of antiplatelet agents (none, aspirin, or clopidogrel), all baseline covariates and discharge medications were adjusted. The proportional hazards assumption was evaluated using plots of the log-log survival function and Schoenfeld residuals.

To adjust for possible allocation bias, subgroup analyses were conducted by age (age <75 vs ≥75 years), sex, hypertension, diabetes, dyslipidemia, use of statins, high bleeding risk (defined according to the Academic Research Consortium for High Bleeding Risk [ARC-HBR]),17 and type of hospital. There were no missing variables in any of the analyses. All analyses were conducted using SAS Enterprise Guide (version 7.1; SAS Institute Inc., Cary, USA) and R (version 4.1.2; R Foundation for Statistical Computing, Vienna, Austria). A 2-tailed P value <.05 was considered statistically significant.

Baseline characteristics, including comorbidities, medication history, and procedural information for the study cohorts before and after propensity score matching, are presented in table 2 of the supplementary data and table 1, respectively. Of the 4657 eligible patients, 3018 were prescribed antiplatelet agents at discharge (antiplatelet therapy group), while 1639 were not (no therapy group). After propensity score matching, 3268 patients were included in the analysis (1634 in each group). Among those receiving antiplatelet therapy, 1281 were treated with aspirin and 353 with clopidogrel (table 3 of the supplementary data).

No significant differences were found in baseline characteristics, comorbidities, or medication history between the antiplatelet therapy and no therapy groups (all standardized mean differences <0.2) (table 1 and table 2 of the supplementary data). Likewise, no significant differences were found in baseline characteristics among the aspirin, clopidogrel, and no therapy groups (table 3 of the supplementary data). The average treatment duration was 458.2 days in antiplatelet therapy group.

During the 1-year follow-up, 235 MACCE events occurred: 115 in the no therapy group and 120 in the antiplatelet therapy group, with incidence rates of 7.0% and 7.3%, respectively (HR, 1.06; 95%CI, 0.82-1.37) (table 2).

During the 5 years of follow-up, MACCE occurred in 342 patients in the no therapy group and in 346 patients in the antiplatelet therapy group with similar incidence rates of 24.7% and 24.8%, respectively (HR, 0.97; 95%CI, 0.84-1.13) (table 2 and figure 2). No significant differences were observed between groups in the incidence rates of all-cause death (8.8% vs 8.5%; HR, 0.89; 95%CI, 0.68-1.16), cardiovascular death (2.2% vs 1.9%; HR=0.87; 95%CI, 0.49-1.53), MI (2.7% vs 2.7%; HR, 0.97; 95%CI, 0.61-1.56), unplanned revascularization (16.3% vs 16.5%; HR, 1.04; 95%CI, 0.86-1.25), or stroke (2.3% vs 2.5%; HR, 0.87; 95%CI, 0.61-1.24) (table 2 and figure 3). The results from the inverse probability treatment weighting analysis were similar to those from the propensity score matched analysis (table 4 of the supplementary data).

Figure 2.

Cumulative incidence of primary efficacy and safety outcomes. The primary efficacy end point was MACCE which was defined as a composite of all-cause death, myocardial infarction, unplanned revascularization, and stroke. The primary safety outcome was any gastrointestinal bleeding. Hazard ratios were adjusted for discharge medication consisting of proton pump inhibitor/H2 blocker. 95%CI, 95% confidence interval; HR, hazard ratio; MACCE, major adverse cardiac and cerebrovascular events.

(0.28MB).

Figure 3.

Cumulative incidence of secondary efficacy outcomes. Kaplan-Meier curves and cumulative incidences of secondary efficacy outcomes between the none vs antiplatelet agent group regarding (A) all-cause death, (B) myocardial infarction, (C) unplanned revascularization, and (D) stroke (including both ischemic and hemorrhagic stroke). Hazard ratios were adjusted for discharge medication consisting of proton pump inhibitor/H2 blocker. 95%CI, 95% confidence interval; HR, hazard ratio.

(0.48MB).

Compared with the no therapy group, neither the aspirin (HR, 0.94; 95%CI, 0.80-1.11) nor the clopidogrel (HR, 0.99; 95%CI, 0.77-1.27) groups showed significant difference in the cumulative incidence of MACCE during the 5 years of follow-up (table 5 of the supplementary data and figure 1 of the supplementary data).

At 1 year, the risk of any bleeding was similar between the antiplatelet therapy and no therapy groups (0.6% vs. 0.6%; HR, 1.11; 95% CI, 0.44-2.83) (table 2), with comparable rates in the aspirin (HR, 1.01; 95% CI, 0.39-2.64) and clopidogrel groups (HR, 0.91; 95%CI, 0.20-4.14) (table 5 of the supplementary data).

However, at 5 years, the antiplatelet therapy group had a significantly higher incidence of gastrointestinal bleeding compared with the no therapy group (1.2% vs 2.2%; HR, 2.07; 95%CI, 1.08-4.00) (table 2 and figure 2). Both the aspirin (HR, 2.03; 95%CI, 1.02-4.05) and clopidogrel groups (HR, 2.56; 95%CI, 0.99-6.64) showed a higher cumulative incidences of gastrointestinal bleeding at 5 years (table 5 of the supplementary data and figure 1 of the supplementary data).

Subgroup analyses showed no apparent trends, except age, favoring the use of antiplatelet agents to prevent MACCE during the 5 years of follow-up. In patients aged ≥75 years, antiplatelet therapy reduced the risk of MACCE compared with no therapy (HR, 0.72; 95%CI, 0.57-0.90), while no benefit was observed in those aged <75 years (HR, 1.09; 95%CI, 0.93-1.29), with a significant interaction (P for interaction <.01) (figure 4). In terms of safety outcomes, the antiplatelet therapy group consistently showed a higher risk of gastrointestinal bleeding among all subgroups (figure 4).

Figure 4.

Subgroup analyses for primary efficacy and safety outcomes at 5 years. Subgroup analyses are shown for (A) MACCE and (B) any gastrointestinal bleeding at 5 years of follow-up. GI, gastrointestinal; HR, hazard ratio; MACCE, major adverse cardiac and cerebrovascular events. *High bleeding risk was defined according to the Academic Research Consortium for High Bleeding Risk (ARC-HBR).

(0.66MB).

Antiplatelet agents have been widely used as part of guideline-directed medical treatment for secondary prevention in patients with established ASCVD.18 In patients undergoing PCI with stent implantation, antiplatelet agents play a pivotal role in preventing stent thrombosis caused by exposed foreign material within the coronary artery until the stent strut is fully re-endothelialized.19 Based on multiple trials showing the benefits of dual antiplatelet therapy immediately after PCI,1–3,20,21 several guidelines recommend it for 6 to 12 months depending on clinical presentation and risk of bleeding, followed by lifelong single antiplatelet therapy.1–3Given that the key pathophysiology of acute coronary syndrome is thrombus formation after plaque rupture, it was expected that antiplatelet therapy would also be effective for primary prevention, especially in patients with cardiovascular risk factors.22 Against this background, numerous trials have been conducted to identify appropriate indications for aspirin in primary prevention. Three representative randomized clinical trials have been published: ASCEND (A Study of Cardiovascular Events in Diabetes), ASPREE (Aspirin in Reducing Events in the Elderly), and ARRIVE (Aspirin to Reduce Risk of Initial Vascular Events). These trials enrolled a total of approximately 47 000 patients with diabetes mellitus, advanced age, or intermediate risk of future ASCVD.8–10 Of these trials, only the ASCEND trial, which enrolled patients with diabetes mellitus, demonstrated a significantly reduced risk of serious vascular events with aspirin compared with placebo. However, in all three trials, including ASCEND, the potential benefits of aspirin were offset by an increased risk of bleeding. Current guidelines, therefore, do not recommend routine aspirin use for primary prevention, advising it only for selected high-ASCVD-risk patients without an excessive bleeding risk.23

Similarly, there have been no specific trials that support the routine use of clopidogrel for primary prevention, although a few studies have evaluated its role as standalone therapy or in combination with aspirin to reduce the risk of ischemic events in different patient populations. In the CAPRIE (Clopidogrel versus Aspirin in Patients at Risk of Ischaemic Events) trial, which enrolled more than 19 000 patients with atherothrombotic events, including recent ischemic stroke, MI, or symptomatic peripheral arterial disease, clopidogrel monotherapy for secondary prevention was superior to aspirin monotherapy in reducing the risk of ischemic events.24 Similar results were also shown by the HOST-EXAM (Aspirin versus clopidogrel for chronic maintenance monotherapy after percutaneous coronary intervention) trial in patients treated with PCI and drug-eluting stent implantation who completed dual antiplatelet therapy for 6 to 18 months after PCI.25 In contrast, the CHARISMA (Clopidogrel for High Atherothrombotic Risk and Ischemic Stabilization, Management, and Avoidance) trial failed to show a benefit of adding clopidogrel to aspirin monotherapy in patients with either clinically evident cardiovascular disease or multiple risk factors.26 In fact, there was a suggestion of harm when the combination therapy was used for primary prevention in high-risk patients with multiple risk factors. These results demonstrate the importance of balancing the potential benefits and harms related to antiplatelet therapy.

The DEFER (Deferral versus performance of percutaneous coronary intervention) trial showed that deferring PCI was superior to performing PCI in patients with functionally insignificant stenosis with FFR >0.75.27 However, this result does not mean that patients with FFR-negative lesions are free from future events. In the FAME 2 trial, the cumulative incidence of death, MI, and any revascularization among patients whose revascularization was deferred based on negative FFR was 4.2%, 8.4%, and 17.5%, respectively, suggesting that these patients do not have a benign prognosis.4 Therefore, they may benefit from aggressive medical treatment. However, current guidelines do not specify how to medically treat these patients with deferred revascularization based on FFR, beyond advising against PCI.1–3In this regard, the current study evaluated the benefits of antiplatelet therapy after deferral of revascularization in patients with established CAD. Despite the presence of intermediate coronary stenosis (50%-70%) on invasive coronary angiography, antiplatelet therapy with either aspirin or clopidogrel did not reduce the risk of death or ischemic outcomes but did increase the risk of gastrointestinal bleeding. These results were consistent among the various subgroups, indicating that they can be generalized to the entire population, regardless of sex, comorbidities, or bleeding risk. A significant interaction was only observed in the subgroup analysis by age: a potential benefit of antiplatelet therapy in patients aged 75 or older may suggest that these patients are probably at an even higher risk of future ischemic events, but this should only be regarded as hypothesis-generating and warrants further study. Overall, the results of the current study suggest that antiplatelet agents should not be routinely prescribed in patients with intermediate coronary artery stenosis after deferral of revascularization based on FFR. Rather, an individualized approach based on the patient's ischemic and bleeding risk should be employed.

The above findings notwithstanding, continued efforts should focus on reducing the risk of future ischemic events in this high-risk population with established CAD. Statins have already been proven to be effective in lowering clinical events in both primary and secondary prevention.28–30 Their protective effect is further supported by imaging studies showing plaque regression and favorable compositional changes with statin therapy.31 Nevertheless, the optimal dosage of lipid-lowering agents in this population remains unclear. In addition, the ongoing HOST-PREVENTION (Harmonizing Optimal Strategy for Treatment of Coronary Artery Stenosis- CloPidogREl for Primary prevention, NCT05845489) trial will further evaluate the role of clopidogrel monotherapy in patients with clinically insignificant coronary artery stenosis. Future studies should also clarify whether identifying high-risk plaque characteristics by invasive or noninvasive imaging could help to stratify patients who would benefit from more aggressive medical treatment, including antiplatelet therapy.

Several limitations of this study should be acknowledged. First, detailed lesion characteristics could not be assessed as the study used an administrative claims database. However, FFR procedures are reimbursed under strict insurance criteria and are routinely audited by the NHIS, providing reasonable assurance that the study population had corresponding lesions and that PCI was not performed in these cases. Second, actual FFR values were unavailable, and we assumed that deferred cases were based on negative FFR. Given that previous multicenter prospective registries in Korea reported deferral rates despite positive FFR of approximately 5.7% (FFR ≤0.75)32 and 9.5% (FFR ≤0.80),7 the influence of deferred revascularization despite positive FFR on these results is likely minimal. Third, although we used propensity score matching, inherent confounding and selection bias could not be completely ruled out. Fourth, despite the large sample size and nationwide cohort data involving over 3000 patients, this study may have been underpowered to detect potential benefits of antiplatelet therapy. Fifth, FFR testing decisions in patients with intermediate stenosis were at the discretion of the treating physician, which could have introduced selection bias.