Abstract
Purpose
Cardiac rehabilitation (CR) is central to tertiary prevention after acute myocardial infarction (AMI), but the prognostic impact of different education and therapy patterns in Asian real world practice is uncertain.
Materials and Methods
Using Korean National Health Insurance Service claims (2018–2022), we identified patients aged ≥40 years hospitalized for AMI who underwent thrombolysis, percutaneous coronary intervention, or coronary artery bypass grafting. CR exposure was classified into six groups by structured education and the number of supervised therapy sessions, and outcomes over 2 years included all cause readmission, readmission and emergency room (ER) visits for MI or unstable angina (UA), all cause mortality, repeat revascularization, and stroke. Multivariable Cox models estimated adjusted hazard ratios (HRs) with 95% confidence intervals (CIs).
Results
Among 92968 patients, 85.5% received no CR, 3.2% had education only, and 11.3% received any therapy. Education only participants had the lowest crude rates of mortality, cardiovascular readmission, and repeat revascularization. All CR exposed groups showed lower all cause mortality, with the greatest benefit in education plus ≥4 therapy sessions (HR 0.257, 95% CI 0.123–0.539) and ≥4 therapy sessions without education (HR 0.291, 95% CI 0.203–0.417). Education focused CR was additionally associated with lower cardiovascular readmission and repeat revascularization, whereas readmission and ER visits varied across therapy intensive groups.
Conclusion
In this nationwide Korean AMI cohort, CR participation, especially with education, was associated with substantial mortality reduction, suggesting structured education as a pragmatic cornerstone of tertiary prevention, with supervised therapy providing incremental benefit when available.
Keywords: Cardiac rehabilitation, myocardial infarction, education, insurance, health
Graphical Abstract
INTRODUCTION
Cardiovascular disease (CVD) has been a global burden, with expectations of an increase due to population growth and aging. Ischemic heart disease (IHD) has a disability-adjusted life year (DALY) rate of 2275.9 per 100000 population worldwide in 2022.1 The situation in Korea is likewise, with a mortality rate of 64.8 per 100000 population and a DALY of 982 per 100000 population in 2020.2
It is important to manage complications and recurrent CV events, and lifestyle changes and guideline-directed medical therapy are essential. However, an emphasis on cardiac rehabilitation (CR) has been implemented in addition to interventional and pharmacological treatments. The first need for CR took shape in 1944, when Cassidy and Campbell3 mentioned in The Lancet the need for rehabilitation due to complications following myocardial infarction (MI). CR is defined as “the coordinated sum of activities required to favorably influence the underlying causes of CVD and to provide the best possible physical, mental, and social conditions, enabling patients, through their own efforts, to preserve or resume optimal functioning within their community and, through improved health behaviors, to slow or reverse the progression of the disease.”4 It is comprised of a comprehensive, patient-focused program that helps not only improve the physical abilities of a patient, but also improves the overall condition of the patient, helping the patient resume day-to-day and social activities despite the remaining risks of another event and the current limitation after an acute CV event.5
CR substantially reduces mortality and hospital readmission rates for a range of cardiac conditions. Landmark studies show that CR after acute MI (AMI) can result in approximately 25% lower hospital readmission rates and 42% lower mortality compared to non-participation.6 After coronary artery bypass grafting (CABG), CR participation has been associated with a 46% reduction in 10-year mortality compared to those who do not participate.7 From a population health perspective, it is therefore important to clarify how the degree of exposure to CR education and therapy relates to subsequent health outcomes in real-world practice, rather than simply whether CR was used or not.
Cohort studies based in regional cardiocerebrovascular centers and other high-volume institutions have prospectively or retrospectively followed patients who participated in structured CR programs to assess clinical, functional, and quality-of-life outcomes.8,9,10 These investigations generally showed favorable effects of CR but were limited by single-center or few-center designs, restricted catchment areas, and potential selection bias, which limited their generalizability to the broader AMI population. Nationwide analyses using health insurance claims data have evaluated associations between CR participation and major outcomes such as mortality and rehospitalization,11,12,13 revealing persistently low participation rates despite clear prognostic benefit. Nonetheless, prior claims-based studies have typically classified CR as a binary exposure and have not fully disentangled the respective contributions of education-focused interventions and varying intensities of supervised therapy.
It is necessary to evaluate how different patterns of CR exposure, such as the presence of structured education and the intensity of supervised therapy sessions, relate to key health outcomes at the national level in Korea. The introduction of National Health Insurance (NHI) reimbursement for CR in 2017 enabled analyses of CR utilization and outcomes using large-scale data. To address this need, this study used nationwide National Health Insurance Service (NHIS) claims data to construct a contemporary AMI cohort and to examine the association between detailed CR exposure patterns, including education and graded levels of therapy, and subsequent outcomes such as all-cause mortality, cardiovascular events, readmissions, repeat revascularization, and stroke over a 2-year follow-up.
MATERIALS AND METHODS
Study design and data source
This retrospective cohort study was conducted using research data provided by the NHIS (NHIS-2025-07-1-115). All Koreans use health insurance provided by the NHI program (about 97%) or the medical aid program (about 3%). The study data from the NHIS included the claimed data submitted by healthcare institutions for services reimbursed under the NHI program and personal data of subscribers and beneficiaries of health insurance. Our study focused on evaluating participation in CR, including education, assessment, and therapy, during hospitalization.
Study population
The selection process of the study subjects is shown in Fig. 1. The initial study population included patients aged 40 years and older who were hospitalized for AMI between January 2018 and December 2022 (n=596297). Eligible patients were operationally defined as those admitted through the emergency department with a primary or secondary diagnosis of AMI, identified using International Classification of Diseases, 10th Revision (ICD-10) codes I21 or I22.
Fig. 1. Flowchart of the study. AMI, acute myocardial infarction; IHD, ischemic heart disease.
Patients were excluded if they had at least one admission or outpatient visit to a clinic-level or higher institution within 1 year prior to the index admission with a diagnosis of angina pectoris (I20), AMI (I21), subsequent MI (I22), complications following AMI (I23), other acute IHD (I24), or chronic IHD (I25) (n=423420). Those who did not receive acute-phase treatment during hospitalization were excluded (n=75092). Acute-phase treatment for AMI included thrombolysis, percutaneous coronary intervention (PCI), or CABG, and the corresponding procedure codes were as follows: thrombolysis (M6634); PCI (M6551, M6552, M6553, M6561, M6562, M6563, M6564, M6565, M6571, M6572, M6638); and CABG (O1640, O1641, O1642, O1647, O1648, O1649, OA640, OA641, OA642, OA647, OA648, OA689). In instances where multiple procedures were claimed, categorization was based on the hierarchy of treatment intensity: CABG was prioritized over PCI, and PCI over thrombolysis. Those who died during hospitalization were excluded (n=4807), and an additional number of patients were excluded due to insufficient data or follow-up period (n=10).
A total of 92968 participants who met the inclusion and exclusion criteria were included in the final analysis. Participants were categorized into six groups based on CR education participation and the level of therapy participation (no participation, 1–3 sessions, and ≥4 sessions). CR education was defined based on insurance reimbursement claims (MM451) for education during hospitalization, and CR therapy frequency was determined by the total number of reimbursement claims (MM453) during hospitalization and up to 1 year after discharge. Group 1 consisted of patients who received neither CR education nor therapy (n=79490). Group 2 consisted of patients who received no education and 1–3 therapy sessions (n=8007). Group 3 consisted of patients who received no education and ≥4 therapy sessions (n=1204). Group 4 consisted of patients who received education only, without therapy (n=2964). Group 5 consisted of patients who received education and 1–3 therapy sessions (n=995). Group 6 consisted of patients who received education and ≥4 therapy sessions (n=308).
Measurement of health outcome variables
As health outcomes of CR, we used six health indicators, comprising three indicators of healthcare utilization and three indicators of disease incidence. The healthcare utilization indicators included all-cause readmission, readmission due to MI or unstable angina (UA), and emergency room (ER) visits due to MI or UA. The disease incidence indicators included all-cause mortality, repeat revascularization due to MI, and stroke incidence. All health outcomes were followed for 2 years from the date of discharge, with CR exposure defined during a 1-year accumulation period from discharge.
The operational definitions of the six health indicators were as follows. All-cause mortality was determined using mortality data provided by Statistics Korea and linked to the NHIS database. All-cause mortality was determined using the date of death, and information on cause-specific mortality was not available in a sufficiently reliable and detailed form. Repeat revascularization due to MI was defined as the performance of an additional revascularization procedure after discharge due to MI and was identified using MI diagnosis codes (I21 and I22) in combination with procedure codes for thrombolysis, PCI, or CABG. Stroke incidence was defined as hospitalization with a diagnosis of stroke after discharge and was identified using claim codes I60–I69.
All-cause readmission was defined as hospitalization for any cause after discharge and was identified using admission claim codes from the ICD. Readmission due to MI or UA was defined as hospitalization for MI or UA after discharge. MI was identified using claim codes I21, I22, I23, I25.2, and I25.5, and UA was identified using claim codes I20.0 and I24. ER visits due to MI or UA were defined as ER visits associated with a diagnosis of MI or UA and were identified using ER visit codes (emergency medical management fee codes; AC101, AC103, AC105, V1100, V1200, V1300, V1400) in combination with the corresponding diagnosis codes.
Measurement of covariates
As covariates, we included sex, age, economic status, place of residence, length of hospital stay (LOS), type of medical institution, method of revascularization, and the Charlson Comorbidity Index (CCI). Age was categorized into five groups: 40–49, 50–59, 60–69, 70–79, and ≥80 years. Economic status was classified into five categories based on the monthly NHI premium level: medical aid beneficiaries, Grade 1, Grade 2, Grade 3, and Grade 4, with higher grades indicating better economic status. Place of residence was categorized into four groups: Seoul, metropolitan cities, urban areas in provinces, and rural areas. LOS was categorized into three groups: 1–7 days, 8–14 days, and ≥15 days. Type of medical institution was categorized into three groups: tertiary hospitals, general hospitals, and hospitals. Methods of revascularization were classified into three categories: CABG, PCI, and thrombolysis. CCI was calculated using values provided by the NHIS and categorized into three groups: 1, 2, and ≥3.
Statistical analysis
Descriptive statistics were calculated to summarize the general characteristics of participants across the six study groups. Subsequently, values for the six health indicators were calculated for each group. Multivariable Cox proportional hazards regression analyses were then performed to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the six health indicators across the five comparison groups, using Group 1 as the reference group after adjustment for covariates. All statistical analyses were conducted using SAS software (version 7.1; SAS Institute, Cary, NC, USA) and R statistical software (version 4.1.3; R Foundation for Statistical Computing, Vienna, Austria).
Ethics statement
The Institutional Review Board of Konkuk University (IRB No. 7001355-202408-E-830) approved the study protocol. The requirement for informed consent was waived due to the retrospective nature of the study. All data obtained from the Korean NHIS were fully anonymized prior to analysis.
RESULTS
General characteristics of the study population
The general characteristics of the study population are presented in Table 1.
Table 1. Baseline Characteristics of the Study Population According to Cardiac Rehabilitation Participation Patterns.
| Characteristics | Total (n=92968) | Group 1 (n=79490) | Group 2 (n=8007) | Group 3 (n=1204) | Group 4 (n=2964) | Group 5 (n=995) | Group 6 (n=308) | p | |
|---|---|---|---|---|---|---|---|---|---|
| Sex | <0.001 | ||||||||
| Male | 71152 (76.5) | 60534 (76.2) | 6338 (79.2) | 963 (80.0) | 2276 (76.8) | 788 (79.2) | 253 (82.1) | ||
| Female | 21816 (23.5) | 18956 (23.8) | 1669 (20.8) | 241 (20.0) | 688 (23.2) | 207 (20.8) | 55 (17.9) | ||
| Age (yr) | <0.001 | ||||||||
| 40–49 | 9511 (10.2) | 7982 (10.0) | 845 (10.6) | 203 (16.9) | 300 (10.1) | 131 (13.2) | 50 (16.2) | ||
| 50–59 | 22016 (23.7) | 18692 (23.5) | 1930 (24.1) | 358 (29.7) | 691 (23.3) | 254 (25.5) | 91 (29.5) | ||
| 60–69 | 26096 (28.1) | 21964 (27.6) | 2461 (30.7) | 428 (35.5) | 841 (28.4) | 293 (29.4) | 109 (35.4) | ||
| 70–79 | 21730 (23.4) | 18726 (23.6) | 1827 (22.8) | 161 (13.4) | 759 (25.6) | 219 (22.0) | 38 (12.3) | ||
| ≥80 | 13615 (14.6) | 12126 (15.3) | 944 (11.8) | 54 (4.5) | 373 (12.6) | 98 (9.8) | 20 (6.5) | ||
| Economic status | <0.001 | ||||||||
| Medical aid beneficiaries | 13841 (14.9) | 12007 (15.1) | 1065 (13.3) | 164 (13.6) | 429 (14.5) | 146 (14.7) | 30 (9.7) | ||
| Grade 1 | 15018 (16.2) | 12931 (16.3) | 1261 (15.7) | 171 (14.2) | 472 (15.9) | 139 (14.0) | 44 (14.3) | ||
| Grade 2 | 15001 (16.1) | 12942 (16.3) | 1214 (15.2) | 163 (13.5) | 465 (15.7) | 173 (17.4) | 44 (14.3) | ||
| Grade 3 | 18184 (19.6) | 15564 (19.6) | 1560 (19.5) | 227 (18.9) | 576 (19.4) | 191 (19.2) | 66 (21.4) | ||
| Grade 4 | 30924 (33.3) | 26046 (32.8) | 2907 (36.3) | 479 (39.8) | 1022 (34.5) | 346 (34.8) | 124 (40.3) | ||
| Place of residence | <0.001 | ||||||||
| Seoul | 13729 (14.8) | 11455 (14.4) | 1755 (21.9) | 273 (22.7) | 116 (3.9) | 96 (9.6) | 34 (11.0) | ||
| Metropolitan cities | 22220 (23.9) | 18612 (23.4) | 1225 (15.3) | 309 (25.7) | 1767 (59.6) | 231 (23.2) | 76 (24.7) | ||
| Urban areas in provinces | 44913 (48.3) | 38634 (48.6) | 4166 (52.0) | 512 (42.5) | 873 (29.5) | 560 (56.3) | 168 (54.5) | ||
| Rural areas | 12106 (13.0) | 10789 (13.6) | 861 (10.8) | 110 (9.1) | 208 (7.0) | 108 (10.9) | 30 (9.7) | ||
| Length of hospital stay | <0.001 | ||||||||
| 1–7 days | 73077 (78.6) | 62169 (78.2) | 6444 (80.5) | 934 (77.6) | 2507 (84.6) | 795 (79.9) | 228 (74.0) | ||
| 8–14 days | 13647 (14.7) | 11890 (15.0) | 1027 (12.8) | 176 (14.6) | 353 (11.9) | 145 (14.6) | 56 (18.2) | ||
| ≥15 days | 6244 (6.7) | 5431 (6.8) | 536 (6.7) | 94 (7.8) | 104 (3.5) | 55 (5.5) | 24 (7.8) | ||
| Method of revascularization | <0.001 | ||||||||
| CABG | 3767 (4.1) | 2606 (3.3) | 981 (12.3) | 88 (7.3) | 46 (1.6) | 37 (3.7) | 9 (2.9) | ||
| PCI | 89194 (95.9) | 76878 (96.7) | 7026 (87.7) | 1115 (92.6) | 2918 (98.4) | 958 (96.3) | 299 (97.1) | ||
| Thrombolysis | 7 (0.0) | 6 (0.0) | 0 (0.0) | 1 (0.1) | 0 (0.0) | 0 (0.0) | 0 (0.0) | ||
| Charlson Comorbidity Index | <0.001 | ||||||||
| 1 point | 8801 (9.5) | 7572 (9.5) | 720 (9.0) | 113 (9.4) | 249 (8.4) | 102 (10.3) | 45 (14.6) | ||
| 2 points | 15497 (16.7) | 13092 (16.5) | 1392 (17.4) | 253 (21.0) | 526 (17.7) | 171 (17.2) | 63 (20.5) | ||
| ≥3 points | 68670 (73.9) | 58826 (74.0) | 5895 (73.6) | 838 (69.6) | 2189 (73.9) | 722 (72.6) | 200 (64.9) | ||
| Type of medical institution | <0.001 | ||||||||
| Tertiary hospital | 36520 (39.3) | 28307 (35.6) | 5114 (63.9) | 640 (53.2) | 2029 (68.5) | 333 (33.5) | 97 (31.5) | ||
| General hospital | 53238 (57.3) | 48297 (60.8) | 2710 (33.8) | 549 (45.6) | 861 (29.0) | 625 (62.8) | 196 (63.6) | ||
| Hospital | 3210 (3.5) | 2886 (3.6) | 183 (2.3) | 15 (1.2) | 74 (2.5) | 37 (3.7) | 15 (4.9) | ||
CABG, coronary artery bypass grafting; PCI, percutaneous coronary intervention.
Data are presented as n (%). Group 1: no education, no therapy; Group 2: no education, 1–3 therapy sessions; Group 3: no education, ≥4 therapy sessions; Group 4: education only; Group 5: education+1–3 therapy sessions; Group 6: education+≥4 therapy sessions.
Among 92968 patients with AMI, 76.5% were male, and the most common age group was 60–69 years, followed by 50–59 years, 70–79 years, ≥80 years, and 40–49 years, with broadly similar sex and age distributions across the six CR groups. Overall, 85.5% did not receive any CR, while 14.5% received CR education and/or therapy.
Regarding economic status, 14.9% were medical aid beneficiaries, and the remainder were distributed across insurance contribution grades 1–4 (16.2%, 16.1%, 19.6%, and 33.3%, respectively), with higher grades modestly enriched in higher-intensity CR groups and medical-aid patients more common in the no-CR group. For residence, 14.8% lived in Seoul, 23.9% in other metropolitan cities, 48.3% in urban provincial areas, and 13.0% in rural areas, with patients from metropolitan regions slightly overrepresented in CR therapy and education-only groups.
Most patients underwent PCI (95.9%), 4.1% had CABG, and <0.1% received thrombolysis, with CABG relatively more frequent in groups receiving ≥4 CR therapy sessions. Comorbidity burden was high: CCI scores of 1, 2, and ≥3 points were observed in 9.5%, 16.7%, and 73.9% of patients, respectively, and the predominance of CCI ≥3 was consistent across CR patterns.
Acute care was provided mainly at general (57.3%) and tertiary (39.3%) hospitals, with higher-intensity CR groups more often treated at tertiary centers. Length of stay was 1–7 days in 78.6% of patients, 8–14 days in 14.7%, and ≥15 days in 6.7%, with only a slightly higher proportion of prolonged stays in intensive CR groups.
Overall, the distribution of demographic, socioeconomic, and clinical characteristics showed statistically significant differences (all p<0.001).
Cumulative incidence rates of six health outcomes
Cumulative incidence rates for 2 years of the six health outcomes are presented in Table 2.
Table 2. Cumulative Incidence Rates of Six Health Outcomes According to Cardiac Rehabilitation Participation Patterns.
| Outcomes | Total (n=92968) | Group 1 (n=79490) | Group 2 (n=8007) | Group 3 (n=1204) | Group 4 (n=2964) | Group 5 (n=995) | Group 6 (n=308) | p |
|---|---|---|---|---|---|---|---|---|
| All-cause readmission | 60360 (64.9) | 51756 (65.1) | 5117 (63.9) | 792 (65.8) | 1738 (58.6) | 729 (73.3) | 228 (74.0) | <0.001 |
| Readmission due to MI or UA | 28220 (30.4) | 24350 (30.6) | 2330 (29.1) | 371 (30.8) | 707 (23.9) | 342 (34.4) | 120 (39.0) | <0.001 |
| ER visits due to MI or UA | 9100 (9.8) | 7642 (9.6) | 840 (10.5) | 161 (13.4) | 333 (11.2) | 92 (9.3) | 32 (10.4) | <0.001 |
| All-cause mortality | 10233 (11.0) | 9214 (11.6) | 651 (8.1) | 30 (2.5) | 264 (8.9) | 67 (6.7) | 7 (2.3) | <0.001 |
| Repeat revascularization due to MI | 12323 (13.3) | 10526 (13.2) | 1257 (15.7) | 143 (11.9) | 235 (7.9) | 119 (12.0) | 43 (14.0) | <0.001 |
| Strokes | 2350 (2.5) | 2026 (2.6) | 201 (2.5) | 27 (2.2) | 67 (2.3) | 27 (2.7) | 2 (0.7) | <0.001 |
MI, myocardial infarction; UA, unstable angina; ER, emergency room.
Data are presented as n (%). Group 1: no education, no therapy; Group 2: no education+1–3 therapy sessions; Group 3: no education, ≥4 therapy sessions; Group 4: education only; Group 5: education+1–3 therapy sessions; Group 6: education+≥4 therapy sessions.
All-cause readmission occurred in 60360 patients (64.9%) overall, ranging from 58.6% in Group 4 to 74.0% in Group 6. Readmission due to MI or UA was observed in 28220 patients (30.4%), with incidences from 23.9% in Group 4 to 39.0% in Group 6, compared with 30.6% in Group 1. ER visits for MI or UA occurred in 9100 patients (9.8%), with group-specific rates between 9.3% and 13.4%. All-cause mortality was 11.0% overall (10233 deaths), lowest in Group 3 (2.5%) and Group 6 (2.3%), compared to Group 1 (11.6%). Repeat revascularization due to MI occurred in 12323 patients (13.3%), ranging from 7.9% in Group 4 to 15.7% in Group 2. Stroke developed in 2350 patients (2.5%), with cumulative incidences varying from 0.7% in Group 6 to about 2.7% in intermediate CR exposure groups, compared with 2.6% in Group 1.
In supplementary analyses of outcome timing, event proportions differed across CR groups but showed broadly consistent temporal trends (Supplementary Fig. 1, only online). Event rates for all outcomes showed a peak within the first month after discharge and fell in the 1–3-month interval, followed by a secondary rise during 3–12 months and a modest plateau or slight decline between 1 and 2 years, with the highest absolute event rates observed in Group 1 and Group 2 for most outcomes.
Association between patterns of CR exposure and health outcome variables
We used multivariable Cox proportional hazards models to evaluate the association between patterns of CR exposure and time-to-event outcomes across the six variables with 95% CIs (Fig. 2).
Fig. 2. Hazard ratios for clinical outcomes according to cardiac rehabilitation education and session exposure after adjusting for the variables of sex, age, economic status, place of residence, length of hospital stay, method of revascularization, Charlson Comorbidity Index, and type of medical institution. Group 2: no education, 1–3 therapy sessions; Group 3: no education, ≥4 therapy sessions; Group 4: education only; Group 5: education+1–3 therapy sessions; Group 6: education+≥4 therapy sessions. MI, myocardial infarction; UA, unstable angina.
For all-cause readmission, risks were generally higher or neutral among therapy-focused groups but not in the education only pattern. Compared with Group 1, HRs for all-cause readmission were 1.031 (1.001–1.062) in Group 2, 1.102 (1.027–1.182) in Group 3, 0.875 (0.834–0.919) in Group 4, 1.168 (1.085–1.256) in Group 5, and 1.228 (1.078–1.398) in Group 6. Education without therapy was associated with fewer readmissions, whereas therapy-only and combined patterns tended to have slightly increased or similar readmission rates.
Readmission due to MI or UA showed a heterogeneous pattern across CR groups. Group 2 had an HR of 0.976 (0.934–1.020) and Group 3 an HR of 1.052 (0.949–1.166), both indicating increased ischemic readmission, whereas Group 4 had a reduced risk with an HR of 0.766 (0.710–0.826); Groups 5 and 6 had higher HRs of 1.115 (1.002–1.241) and 1.299 (1.085–1.554), respectively.
For ER visits due to MI or UA, only Group 2 showed an elevated risk. HRs were 0.983 (0.913–1.058) for Group 2, 1.475 (1.264–1.726) for Group 3, 1.129 (1.009–1.263) for Group 4, 0.989 (0.805–1.215) for Group 5, and 1.217 (0.860–1.723) for Group 6, indicating that more intensive therapy without concurrent education was associated with greater unscheduled ER utilization, whereas education-oriented patterns were neutral.
For all-cause mortality, HRs were 0.726 (0.669–0.787) for Group 2, 0.291 (0.203–0.417) for Group 3, 0.803 (0.709–0.909) for Group 4, 0.657 (0.517–0.836) for Group 5, and 0.257 (0.123–0.539) for Group 6, indicating the lowest mortality in Groups 3 and 6.
Repeat revascularization due to MI was more frequent in lower-intensity therapy groups but appeared attenuated or neutral in higher-intensity or education-focused patterns. HRs were 1.160 (1.087–1.227) for Group 2, 0.872 (0.739–1.029) for Group 3, 0.597 (0.524–0.680) for Group 4, 0.867 (0.723–1.038) for Group 5, and 1.054 (0.739–1.422) for Group 6, highlighting a markedly lower revascularization risk in the education-only group and a modest increase in those with limited therapy exposure plus education.
Stroke incidence was not significantly modified by any CR participation pattern. HRs for stroke were 0.967 (0.833–1.122) in Group 2, 1.075 (0.734–1.573) in Group 3, 0.939 (0.733–1.203) in Group 4, 1.176 (0.080–1.720) in Group 5, and 0.322 (0.126–1.287) in Group 6.
DISCUSSION
This nationwide, population-based cohort analysis aimed to demonstrate the impact of CR education and therapy on post-AMI outcomes within a real-world Korean setting. This study extended previous Korean big-data analyses11,12,13 by disentangling the prognostic contributions of education-only, combined education and therapy, and therapy-only patterns over a uniform 2-year follow-up period.
The graded reduction in all-cause mortality with increasing therapy sessions indicates a possible dose-response relationship between CR adherence and clinical benefit. This pattern is consistent with prior randomized trials and meta-analyses of comprehensive CR programs, which have generally shown the strongest and most consistent effects of CR on mortality, whereas reductions in non-fatal events are smaller or less uniform.6,7,12,14,15,16 In addition to the direct effects of supervised exercise, risk-factor management, and structured education, patients who engage with CR may also have more favorable unmeasured characteristics such as higher motivation, better health literacy, and greater socioeconomic resources. Also, patients perceived by clinicians as more fragile or at higher risk, such as those with complex coronary disease or heart failure, may have been more strongly encouraged or referred to intensive, center-based CR. Taken together, these findings may reflect both the intrinsic benefits of CR and a selection effect.
Our findings also show that patients receiving CR education without therapy tended to have lower risks of all-cause mortality, hospital readmission, and repeat revascularization due to MI and UA. In particular, the favorable outcomes of the education-only group suggest that early, structured educational interventions during hospitalization or shortly after discharge may serve as a pragmatic and scalable strategy in systems where supervised exercise programs are underdeveloped or unevenly distributed.13,17 Education-driven improvements in health literacy, self-management, and adherence to guideline-directed medical therapy may yield substantial prognostic benefit even in the absence of intensive supervised exercise.
From the perspective of healthcare utilization, the heterogeneous effects we observed for all-cause readmission and ER visits suggest patterns that differ from those seen for outcomes such as mortality or recurrent MI. In particular, ER visits and readmission for MI or UA may reflect differences in baseline functional status, comorbidity burden, or healthcare-seeking behavior among patients who require more intensive CR therapy sessions, rather than purely differences in disease incidence. Extending beyond simple between-group differences, these findings also imply that increasing the number of supervised sessions alone may be insufficient to optimize post-AMI care without concurrent standardization of program components, rigorous risk-factor control, and individualized behavioral and psychosocial interventions.
Interestingly, participation in exercise-based CR was associated with higher risks of all-cause readmission and readmission for MI or UA, with a dose–response relationship according to the number of supervised sessions. Rather than implying harm from CR, this pattern likely reflects confounding by indication and residual differences in underlying vulnerability. Also, frailty and related geriatric syndromes are well-established predictors of readmission and adverse outcomes after cardiovascular events, which may have contributed to the outcomes. Unfortunately, due to the nature of administrative data, it is difficult to capture the details of these patients.
By contrast, stroke incidence in our cohort showed a pattern distinct from both mortality and recurrent MI or UA, with no clear association with CR exposure. This divergence underscores that cerebrovascular events may be driven more strongly by long-term comorbidity profiles and systemic vascular risk than by the CR participation patterns that influence ischemic cardiac events and healthcare utilization. The absence of clear benefit for stroke and ER visits is consistent with prior work,8 suggesting that cerebrovascular events and unscheduled care use are strongly influenced by broader comorbidity profiles, frailty, and social determinants that may not be fully modifiable within standard CR frameworks.
International data have repeatedly demonstrated mortality and rehospitalization benefits of CR after AMI and coronary revascularization,6,7,18 and our study confirms these benefits in an East Asian population using a comprehensive national insurance claims database. Korean studies have documented low CR uptake and substantial regional and institutional disparities in program availability, staffing, and infrastructure.11,12,13,18 Our outcome-based findings provide additional support for national strategies aimed at expanding and upgrading CR services. Recent Korean initiatives to enhance CR capacity within regional cardiocerebrovascular centers and refine reimbursement schemes are in line with our observation that education-focused interventions deliver meaningful benefit even where full exercise-based programs remain constrained.8,11,13
From a health-system perspective, education-centered CR could be a highly cost-effective intervention that can reduce mortality and CV events with relatively modest resource requirements compared with fully supervised exercise programs. Our findings suggest that using standardized educational bundles, such as addressing pharmacotherapy, risk-factor modification, physical activity guidance, diet, and symptom recognition, could be considered a minimum quality standard of post-AMI care across all hospitals. Moreover, the favorable outcomes in patients who combined education with multiple therapy sessions support current guideline recommendations advocating comprehensive, multidisciplinary CR for all eligible patients whenever feasible. Our analysis classified CR exposure and education into six types of patterns rather than modeling education and therapy as separate variables. As a result, we cannot fully clarify whether the observed prognostic differences are predominantly resulting from structured education, supervised exercise, or their interaction. Future studies using more flexible modeling approaches will be needed to clarify the distinct and synergistic contributions of these CR components.
This study has the following limitations. This study is a retrospective cohort analysis based on administrative claims data and is therefore subject to residual confounding and selection bias despite multivariable adjustment. Patients who participate in CR, especially those who completed more frequent therapy, may have different baseline risk profiles, health behaviors, or motivation levels than non-participants, and such factors were not fully portrayed with this data. Also, patients in higher-intensity groups had to survive long enough to complete multiple sessions, whereas early deaths were necessarily assigned to lower-exposure or non-CR groups. This exposure definition makes our design prone to immortal time bias, which would be expected to overestimate the apparent survival benefit of high-session CR groups. In particular, the very low HRs for all-cause mortality observed in Group 3 (HR 0.291) and Group 6 (HR 0.257) should be interpreted with caution, because the true mortality reduction associated with intensive CR participation is likely more modest than these point estimates suggest.
We were unable to adjust for use and adherence to key secondary-prevention medications (e.g., statins, beta-blockers, antiplatelet agents) due to the limitations in obtaining such data. Differences from evidence-based pharmacotherapy and medication adherence, therefore, may have contributed to the observed prognostic differences between CR patterns. This is an important issue that would necessitate further research in the future.
The content, intensity, and quality of CR education and therapy were not directly assessed, and inter-hospital variability may have attenuated or obscured the true effects of optimally delivered programs. Additionally, misclassification is possible because informal counseling, home-based exercise, or non-claim–based educational activities could not be identified within the claims structure. Outcomes were restricted to events recorded in the NHIS and did not include patient-reported metrics such as quality of life, functional capacity, or psychosocial status, which are important targets of CR. Finally, the study was conducted within the Korean universal coverage system and current CR reimbursement policies, which may limit generalizability to countries with different financing structures or patterns of CR delivery.
This nationwide cohort study demonstrates that participation in CR after AMI is associated with substantial reductions in all-cause mortality in real-world Korean practice. Education-focused CR may represent a pragmatic strategy in settings that lack adequate CR personnel or facilities, supporting the prioritization of early, standardized educational interventions for all eligible AMI patients, alongside strategic expansion of supervised CR capacity and infrastructure to reduce the persistent implementation gap in Korea.
Future research should clarify the optimal content, timing, and delivery modalities of CR education, including digital and hybrid models, and identify patient subgroups that derive the greatest incremental benefit from higher-intensity therapy. In addition, implementation studies are needed to address barriers to CR referral, enrollment, and completion at the patient, provider, and institutional levels, particularly in under-resourced regions and smaller hospitals. Strengthening national CR registries and linking them with clinical and claims data will be critical to monitor quality, benchmark performance, and guide iterative improvements in CR-related policy and practice, while incorporating patient-centered outcomes to more fully capture the multidimensional benefits of comprehensive CR.
ACKNOWLEDGEMENTS
This research was supported by Konkuk University in 2025.
Footnotes
The authors have no potential conflicts of interest to disclose.
- Conceptualization: Kina Jeon, Hyeongsu Kim, and Kyeong Eun Uhm.
- Data curation: Hyeongsu Kim, Chul Kim, Kunsei Lee, Bora Lee, and Ho Jin Jeong.
- Formal analysis: Kina Jeon, Hyeongsu Kim, Chul Kim, Kunsei Lee, Bora Lee, and Ho Jin Jeong.
- Funding acquisition: Hyeongsu Kim.
- Investigation: Kina Jeon, Hyeongsu Kim, Sung Hea Kim, and Kyeong Eun Uhm.
- Methodology: Kina Jeon, Hyeongsu Kim, Sung Hea Kim, and Kyeong Eun Uhm.
- Project administration: Kina Jeon, Hyeongsu Kim, Sung Hea Kim, and Kyeong Eun Uhm.
- Resources: Hyeongsu Kim, Chul Kim, Kunsei Lee, Bora Lee, and Ho Jin Jeong.
- Software: Hyeongsu Kim, Chul Kim, Kunsei Lee, Bora Lee, and Ho Jin Jeong.
- Supervision: all authors.
- Validation: all authors.
- Visualization: Kina Jeon, Hyeongsu Kim, and Kyeong Eun Uhm.
- Writing—original draft: Kina Jeon and Hyeongsu Kim.
- Writing—review & editing: Kina Jeon and Hyeongsu Kim.
- Approval of final manuscript: all authors.
SUPPLEMENTARY MATERIAL
Outcome event cases per 1,000 persons according to cardiac rehabilitation participation pattern.
References
- 1.Mensah GA, Fuster V, Murray CJL, Roth GA. Global burden of cardiovascular diseases and risks, 1990-2022. J Am Coll Cardiol. 2023;82:2350–2473. doi: 10.1016/j.jacc.2023.11.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Jung YS, Kim YE, Ock M, Yoon SJ. Measuring the burden of disease in Korea using disability-adjusted life years (2008–2020) J Korean Med Sci. 2024;39:e67. doi: 10.3346/jkms.2024.39.e67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Cassidy M, Campbell M. Rehabilitation of cardiac patients. Lancet. 1944;243:484 [Google Scholar]
- 4.Cowie A, Buckley J, Doherty P, Furze G, Hayward J, Hinton S, et al. Standards and core components for cardiovascular disease prevention and rehabilitation. Heart. 2019;105:510–515. doi: 10.1136/heartjnl-2018-314206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Abreu A, Frederix I, Dendale P, Janssen A, Doherty P, Piepoli MF, et al. Standardization and quality improvement of secondary prevention through cardiovascular rehabilitation programmes in Europe: the avenue towards EAPC accreditation programme: a position statement of the Secondary Prevention and Rehabilitation Section of the European Association of Preventive Cardiology (EAPC) Eur J Prev Cardiol. 2021;28:496–509. doi: 10.1177/2047487320924912. [DOI] [PubMed] [Google Scholar]
- 6.Dunlay SM, Pack QR, Thomas RJ, Killian JM, Roger VL. Participation in cardiac rehabilitation, readmissions, and death after acute myocardial infarction. Am J Med. 2014;127:538–546. doi: 10.1016/j.amjmed.2014.02.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Pack QR, Goel K, Lahr BD, Greason KL, Squires RW, Lopez-Jimenez F, et al. Participation in cardiac rehabilitation and survival after coronary artery bypass graft surgery: a community-based study. Circulation. 2013;128:590–597. doi: 10.1161/CIRCULATIONAHA.112.001365. [DOI] [PubMed] [Google Scholar]
- 8.Kim C, Sung J, Han JY, Jee S, Lee JW, Lee JH, et al. Current status of cardiac rehabilitation in the regional cardiocerebrovascular centers in Korea. J Clin Med. 2021;10:5079. doi: 10.3390/jcm10215079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kim KH, Han JY. Fact sheet on cardiac rehabilitation for cardiovascular disease in South Korea. Ann Rehabil Med. 2023;47:318–325. doi: 10.5535/arm.23050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kim C, Sung J, Lee JH, Kim WS, Lee GJ, Jee S, et al. Clinical practice guideline for cardiac rehabilitation in Korea: recommendations for cardiac rehabilitation and secondary prevention after acute coronary syndrome. Korean Circ J. 2019;49:1066–1111. doi: 10.4070/kcj.2019.0194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kim H, Yun RY, Kim C, Shin YI, Bang HJ, Uhm KE, et al. Trends in cardiac rehabilitation participation in patients with acute myocardial infarction: a 5-year nationwide study in Korea. J Korean Med Sci. 2026;41:e8. doi: 10.3346/jkms.2026.41.e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Choi HE, Kim C, Lee DJ, Joo JE, Kim HS. Participation and prognostic impact of cardiac rehabilitation after acute coronary syndrome: big-data study of the Korean National Health Insurance Service. J Korean Med Sci. 2023;38:e119. doi: 10.3346/jkms.2023.38.e119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kim SH, Ro JS, Kim Y, Leigh JH, Kim WS. Underutilization of hospital-based cardiac rehabilitation after acute myocardial infarction in Korea. J Korean Med Sci. 2020;35:e262. doi: 10.3346/jkms.2020.35.e262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Adachi T, Iritani N, Kamiya K, Iwatsu K, Kamisaka K, Iida Y, et al. Prognostic effects of cardiac rehabilitation in patients with heart failure (from a multicenter prospective cohort study) Am J Cardiol. 2022;164:79–85. doi: 10.1016/j.amjcard.2021.10.038. [DOI] [PubMed] [Google Scholar]
- 15.Goel K, Lennon RJ, Tilbury RT, Squires RW, Thomas RJ. Impact of cardiac rehabilitation on mortality and cardiovascular events after percutaneous coronary intervention in the community. Circulation. 2011;123:2344–2352. doi: 10.1161/CIRCULATIONAHA.110.983536. [DOI] [PubMed] [Google Scholar]
- 16.Eijsvogels TMH, Maessen MFH, Bakker EA, Meindersma EP, van Gorp N, Pijnenburg N, et al. Association of cardiac rehabilitation with all-cause mortality among patients with cardiovascular disease in the Netherlands. JAMA Netw Open. 2020;3:e2011686. doi: 10.1001/jamanetworkopen.2020.11686. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Turk-Adawi K, Sarrafzadegan N, Grace SL. Global availability of cardiac rehabilitation. Nat Rev Cardiol. 2014;11:586–596. doi: 10.1038/nrcardio.2014.98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Claes J, Kuznetsova T, Cauwenberghs N, Cornelissen V. The changing landscape of cardiac rehabilitation and the power of personalized therapy. Front Cardiovasc Med. 2024;11:1393217. doi: 10.3389/fcvm.2024.1393217. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Outcome event cases per 1,000 persons according to cardiac rehabilitation participation pattern.