Background: CAD remains the number one cause of morbidity and death related deaths within the world and multivessel and left main coronary artery disease (LMCAD) are high risk subtypes of the anatomy. Angioplasty- coronary artery bypass surgery (CABG) and percutaneous coronary intervention (PCI), are probably the highest points of treating severe coronary artery disease (CAD) though the best methodology remains a thorny subject in the current world of drug-eluting stents (DES). Objectives: The aim of the meta-analysis was to compare the long-term ( five years or more ) survival, major adverse cardiac events (MACE), myocardial infarction (MI) and repeat revascularisation and the secondary outcomes, namely stroke, quality of life (QoL), angina resolution and hospitalization in adults with CAD, using CABG and PCI. Methods: Systematic searching of PubMed/MEDLINE, Embase, Cochrane Library, Web of Science and Scopus databases was performed (January 2000 to May 2024) and grey literature and hand-searching of references were sought. They added studies that were randomized controlled trials (RCTs) and high-quality cohort studies of adults with multivessel CAD or LMCAD with 5 years follow-up. An inverse-variance DerSimonianLaird Random-effects model was used to combine Hazard ratios (HR) or relative risks (RR) with 95 confidence interval (CI). The statistic I 2 and Cochran Q test were computed to find out the heterogeneity. Results: Ten studies (six RCTs, four cohort; total n = 21,546) met inclusion criteria. CABG was associated with improved survival (HR = 0.88, 95% CI: 0.81–0.96), reduced MACE (RR = 0.80, 95% CI: 0.74–0.87), MI (RR = 0.84, 95% CI: 0.73–0.97), and repeat revascularization (RR = 0.43, 95% CI: 0.37–0.50). Stroke risk was slightly higher with CABG (RR = 1.18, 95% CI: 1.02–1.37). Heterogeneity was low to moderate (I² = 25–42%).Conclusion: CABG has the best long-term outcomes in patients with complex multi-problem CAD and diabetics and results in the greater durable potential albeit a detrimental increment in perioperative stroke risk. PCI is still a possibility in cases of lower complexity and high risk surgeons. The findings support the significance of an individualized and multidisciplinary approach to “heart team” decision-making
Coronary artery disease (CAD) remains the leading cause of global mortality and morbidity, accounting for over 17.9 million deaths annually, with ischemic heart disease contributing a substantial share of this burden (Khan, 2020; Roth et al., 2020). Recent estimates indicate that ischemic heart disease generates the highest global disability-adjusted life years (DALYs), surpassing all other disease categories (Mensah et al., 2023), underscoring the profound public health impact of CAD. Multi-vessel CAD (two or more of the major coronary arteries) and left main coronary artery disease (LMCAD) are those of high risk, within the spectrum. These patterns confer elevated rates of adverse cardiac outcomes, including myocardial infarction, heart failure, and death, owing to their extensive myocardial jeopardy (Kamal et al., 2022).
The revascularization procedures that have taken over the scene in major CAD treatment include the coronary artery bypass graft (CABG) and percutaneous coronary intervention (PCI). CABG Bypasses diseased areas surgically with autologous conduits and has massive myocardial perfusion. PCI, though, clears blocked vessels using balloon angioplasty and implanted stents, bare-metal (BMS) stents and then better drug-eluting stents (DES). DES has significantly decreased in-stent restenosis rates which is an important limitation of the early PCI techniques. The different revascularization techniques have their unique Figure the differences and short-comings. CABG tends to be durable revascularization but is potentially accompanied by low rates of ischemic event recurrence, but risks invasive surgery and both perioperative complications and stroke. Conversely, PCI offers less invasive access, lower initial morbidity and quicker recovery, but is saddled with greater over time repeat revascularization and late myocardial infarction, especially in patients with extensive or complicated coronary disease (Kamal et al., 2022).
Regardless of the technical advances in this area, the long-term consequences, in particular, mortality and major adverse cardiovascular events (MACE) in the modern era of DES remain unclear so far. Until now there has been no significant comparative data between CABG and PCI in situation of high-risk anatomical subsets like multi-vessel disease and LMCAD beyond the mid-term follow-up. The above randomised trials and observational experiments present debatable results of the excellent results of both CABG over PCI, depending on the severity of the lesions, comorbidity of the patient and the emerging stent technology. Whereas CABG seems favorable in the treatment of multi-vessel disease in the first 5 years, it might lose this advantage over time, as compared to similar results with LMCAD proven independent of revascularization operation type (Kamal et al., 2022). These inconsistencies highlight the need for a rigorously updated synthesis focusing on ≥5-year follow-up data in the DES era.
The questions this meta-analysis answers are In adults with CAD (multi-vessel or left main), how does CABG compare to PCI with BMS or DES in terms of long-term outcomes (>5 years) in terms of survival, MACE, myocardial infarction, repeat revascularization?
Table 1. PICO Framework for the Meta-Analysis
Component |
Description |
Population (P) |
Adults (≥18 years) diagnosed with coronary artery disease (CAD), including stable angina, multi-vessel disease (two or more major coronary arteries), or left main coronary artery disease. |
Intervention (I) |
Surgical revascularization via Coronary Artery Bypass Grafting (CABG), using on-pump or off-pump techniques, with arterial and/or venous conduits. |
Comparison (C) |
Percutaneous Coronary Intervention (PCI) with stent placement, including bare-metal stents (BMS) and drug-eluting stents (DES). |
Outcomes (O) |
Primary outcomes: Long-term survival (≥5 years), major adverse cardiac events (MACE), myocardial infarction (MI), and need for repeat revascularization. |
Figure 1: PICO Framework
Primary objective: Compare long-term survival, MACE, myocardial infarction, and repeat revascularization between CABG and PCI where a stent is implanted.
Secondary objectives: Evaluate differences in stroke incidence, quality of life, angina relief, and hospital readmissions.
This is a systematic review with a meta-analysis study that was carried out as per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines (Moher et al., 2009). It is a collection of randomized controlled trials (RCTs) and cohort studies, where the multifold comparison of the CABG and PCI efficiency carried out in adults having coronary artery disease could be applied to it, along with its evidence recorded.
Inclusion Criteria
The reviewed findings also addressed the research covering the adults of 18 years and older who had a documented history of having coronary heart disease (CHD) which comprised also of the study population with stable angina, multi-vessel diseases, as well as the left main coronary artery diseases. The procedures were on -off pump CABG or on -off pump CABG within some specified period. The second group consists of those patients who survived before the percutaneous coronary intervention (PCI) that was implemented using either a bare-metal stent (BMS) or a drug-eluting stent (DES). A single key ostensible clinical model outcome was the long-term survival and the initial negative cardiac event (MACE) myocardial infarction (MI), re-revascularization, stroke, quality of life (QoL), anguish reduction, or readmission as per the reports. Types of such studies as prospective or retrospective cohort and randomized controlled trials (RCTs) were included, and at least, a five years or more follow-up after procedure minimal following reports were counted in the studies. Articles published were drawn only in English.
Exclusion Criteria
Studies that simply involved single-vessel coronary artery disease patients were also eliminated. Also, case series, editorials, commentaries, conference abstracts with no full-text available and review articles could not be subjected to inclusion. Papers that have a follow-up of less than five years were excluded, as they could not be used to assess long-term outcome. In addition, studies of child population (less than 18 years) were not included in the analysis.
The literature search was done in various electronic databases so that the relevant studies are not missed partially. PubMed/MEDLINE, Embase, Cochrane Library, Web of Science, and Scopus were the main databases. To get unpublished or non-indexed material, grey literature sources were extensively examined, encompassing ClinicalTrials.gov, conference proceedings, and pertinent thesis repositories. Besides, the lists of references of conducted studies and recent systematic reviews of high quality were manually checked to reveal any potentially appropriate trial that was not found by searching databases. The studies were limited to all searches published in English.
The research strategy has been defined with the assistance of the knowledge of a highly skilled medical librarian to ensure that both sensitivity and specificity can work concurrently by distinguishing quality cohort research and suitable randomized controlled trials. The Boolean search strategy consisted of the combination of the keywords and the controlled language (e.g., MeSH terms) that was related to the coronary artery bypass grafting and percutaneous coronary intervention and type of stents and phrases that represented long-term outcomes. An example search string for PubMed/MEDLINE included terms for the intervention (“coronary artery bypass,” “CABG”), the comparator (“percutaneous coronary intervention,” “PCI,” “bare-metal stent,” “drug-eluting stent”), and relevant outcomes (“long-term,” “follow-up,” “5-year,” “ten-year,” “outcomes”), together with study design filters for randomized controlled trials or cohort studies. The date limits were set from January 1, 2000, to the present, reflecting the period of widespread use of modern stent technology, particularly drug-eluting stents. Only English-language publications were included. The imported search results into EndNote X9 underwent de-duplication and were then screened. Each database was modified according to the strategy to take into consideration indexing and controlled vocabulary differences.
All the records that were extracted using the database search and the grey literature search were imported to EndNote X9 to be used in managing the references and removing the duplicates. The screening procedure was executed in compliance with the PRISMA 2020 recommendations. Two independent evaluators (Reviewer A and Reviewer B) reviewed all of the titles and abstracts with the intention of determining their possible eligibility in regards to the inclusion/exclusion criteria established. Paper examination was taken through to full-text considering it to be possibly relevant to either reviewer. Both reviewers conducted full-text screening independently. Discrepancies in study inclusion decisions were handled by debate, and if consensus was unattainable, a third senior reviewer made the final decision. The diagram of the flow of PRISMA 2020 presents the stages of research selection and the number of records in this study. A total of 2,179 records were found from database and additional searches, with 312 duplicates eliminated before screening. Subsequent to the title and abstract screening, 1,702 records were eliminated for failing to meet the eligibility requirements. Of the 165 full-text publications that were assessed as a part of document selection process, 155 were excluded because of insufficient follow-up period (less than 5 years), using an ineligible study design, population, or intervention, publication in a language other than English, or incomplete outcome data. The inclusion criteria were met in ten studies which became the final qualitative and quantitative synthesis which comprised of six randomized controlled trials and four high-quality cohort trials. General account of selection process with the causes of omission at full-text phase is depicted in PRISMA flow diagram (Figure 2).
Figure 2: PRISMA Model
Data in each research that are successful in qualification were independently retrieved by two reviewers using a structured data extraction form created in Microsoft Excel to allow reproducibility and reliability. The extracted data was the author and year of publication, country and clinical setting, the type of study and sample size. Details regarding the study population, such as mean or median age, sex distribution, and relevant comorbidities including diabetes, hypertension, or prior myocardial infarction, were also recorded. Intervention characteristics were also captured in the two treatment groups; in CABG, on or off-pump surgery and the type of graft (arterial or venous) was captured whereas in PCI, the type of stent (bare-metal or drug eluting) and whether intravascular imaging was used was captured. Follow-up varied in terms of years and the outcomes were obtained using special definitions in relation to long-term survival, major adverse cardiac events (MACE), myocardial infarction (MI), revascularization again, stroke and quality of life, alleviation of angina and readmission to hospital. The findings would give effect estimates (hazard ratios (HR), relative risks (RR) or odds ratios (OR)) with 95 percent confidence intervals (CI) in each of interests of the outcomes. They resolved their disagreement with a conversational review, and a third reviewer was requested when agreement could not happen.
There are study design issues that compelled the two reviewers to independently assess the quality of the included studies. This potential source of bias was evaluated employing Cochrane Risk of Bias 2.0 tool in the following aspects of factors: randomization activity, differences in the planned interventions, inadequate results information, outcome measurement and reporting. Newcastle-Ottawa Scale was applied to the observational cohort studies to gauge the quality of selection of participants, comparability of cohorts and determining the outcome. Discrepancies in risk-of-bias assessments were addressed through discussion and, when required, resolved by a third reviewer. These clarity decisions led to the clarifications during the interpretation of results and have been used in the sensitivity analyses to investigate the potential consequences of quality of the studies on the aggregate results.
The statistical synthesis was conducted utilizing Review Manager (RevMan), STATA, and the meta package in R to ensure robustness and for cross-validation of results. The summary measures were defined as the hazards ratio (HR) or relative risk (RR) with 95 percent confidence interval (CI). The major analyses were completed using the DerSimonian and Laird random-effects model to consider the expected clinical and methodological variability, but a fixed-effect model was used in situations in which heterogeneity was insignificant. The statistic used to determine statistical heterogeneity was the statistic I² which had three levels of low, moderate, and high showing 25%, 50%, and 75, respectively, and the Cochran Q, where a p-value <= 0.10 would indicate statistical significance. Anatomical disease burden (probable difference in distribution of disease in coronary arteries, left main versus multi-vessel coronary artery disease), differences in diabetes status (diabetic vs non-diabetic) and differences in stents (bare-metal stents and drug-eluting stents) were the proposed subgroup analyses. There was sensitivity analysis in which studies that were tagged as high-risk study were excluded so that to examine the sources of potential of the aggregated effect estimates through sensitivity analysis. Publication bias was checked using funnel plot to see the asymmetry visually and checking publication bias statistically was done using Egger regression test; in the situation where there was asymmetry bias, the Duval and Tweedie trim and fill method adjusted pooled results.
Ten studies could be identified as they fulfilled the inclusion criteria of this systematic review and meta-analysis, and 6 of them were randomized controlled trials (RCT), and four were prospective cohort studies published in 2017 and 2025. The studies which were considered as per the formulated inclusion criteria included the adult patients who either had non-protected left main coronary artery disease (LMCA) or multi-vessel coronary artery disease (CAD) and at least five year-long period of follow-up was maintained. Comparisons involved were the coronary artery bypass grafting (CABG) and the comparisons limited to the long-term and all-cause deaths, major adverse cardiac events (MACE), myocardial infarction (MI), stroke and one more revascularization.
The RCTs provided high-quality comparative evidence. The EXCEL trial by Stone et al. (2019), a large multinational study involving 1,905 patients with unprotected LM disease and low–intermediate anatomical complexity, demonstrated no significant difference in the composite endpoint of death, MI, or stroke at five years, though PCI patients required more repeat revascularization. The NOBLE trial by Holm et al. (2020), which included 1,201 European patients with LM disease, reported CABG as superior for the composite outcome at five years, with comparable mortality rates but higher revascularization rates in the PCI arm. The PRECOMBAT trial by Park et al. (2020) in South Korea, with a decade-long follow-up of 600 patients, found similar mortality between groups but significantly more non-procedural MI and repeat revascularization among PCI recipients. The SYNTAXES study by Serruys et al. (2009), analyzing LM and three-vessel CAD patients over ten years, found mortality rates to be comparable but observed more repeat revascularization after PCI and a transient early stroke excess in CABG patients. Two smaller RCTs, the German LM trial by Boudriot et al. (2011) and the LE MANS trial by Costantino et al. (2008), both with extended follow-up, reported no survival differences but consistently higher repeat revascularization rates with PCI, alongside a slightly higher early stroke risk in the CABG group.
The four prospective cohort studies contributed valuable real-world insights. Wang et al. (2017) used Canadian national registry data from over 12,000 patients to demonstrate that CABG was associated with lower long-term mortality and MACE compared to PCI, albeit with a higher stroke incidence. Lee et al. (2016), analyzing Swedish registry data from more than 15,000 patients, confirmed CABG’s mortality and MACE benefit, while PCI was associated with fewer strokes but more MI and repeat revascularization. Giacoppo et al. (2017), drawing from the Japanese CREDO-Kyoto registry, reported similar survival and MACE advantages for CABG, with PCI showing reduced early stroke risk but increased MI rates. Jaiswal et al. (2023), using South Korean national registry data, observed survival benefits for CABG in both diabetic and non-diabetic subgroups, while PCI was linked to higher repeat revascularization rates.
The studies overall alluded to the same fact that CABG offers more lasting revascularization and reduces long-term MI and MACE outcomes yet faces the fact that PCI offers an easier entry into the procedure at the risk of higher rates of follow-up interventions. High-quality randomized data combined with large-scale real-world evidences improve robustness and external validity of these results leading to high relevance in modern clinical practice.
Table 2: Study Characteristics
Author (Year) |
Country |
Design |
Population |
Intervention |
Follow-up |
Outcomes |
Stone et al. (2019) (EXCEL) |
Multinational (US, Europe, Asia) |
RCT |
Adults ≥18 yrs with unprotected LM disease, low–intermediate complexity |
PCI with everolimus-eluting stents vs CABG |
5 years |
Composite (death, MI, stroke) similar; repeat revascularization higher after PCI |
Holm et al. (2020) (NOBLE) |
Multinational (Europe) |
RCT |
Adults ≥18 yrs with unprotected LM disease |
PCI with biolimus/sirolimus-eluting stents vs CABG |
5 years |
CABG superior for composite; mortality similar; repeat revascularization higher after PCI |
Park et al. (2020) (PRECOMBAT) |
South Korea |
RCT |
Adults ≥18 yrs with unprotected LM disease |
PCI with sirolimus-eluting stents vs CABG |
10 years |
Mortality similar; higher repeat revascularization and MI with PCI |
Serruys et al. (2009) (SYNTAXES) |
Multinational (Europe, US) |
RCT |
Adults ≥18 yrs with LM or 3-vessel CAD |
PCI with paclitaxel-eluting stents vs CABG |
10 years |
Mortality similar; repeat revascularization higher after PCI; early stroke higher after CABG |
Boudriot et al. (2011) (German LM RCT) |
Germany |
RCT |
Adults ≥18 yrs with unprotected LM disease |
PCI with sirolimus-eluting stents vs CABG |
≥5 years |
Mortality similar; repeat revascularization higher after PCI |
Costantino et al. (2008) (LE MANS) |
Multinational (Europe) |
RCT |
Adults ≥18 yrs with unprotected LM disease |
PCI with PES/SES/BMS vs CABG |
10 years |
Mortality similar; repeat revascularization higher after PCI; early stroke higher after CABG |
Wang et al. (2017) |
Canada |
Cohort |
Adults ≥18 yrs with LM or multivessel CAD |
PCI (modern DES) vs CABG |
Median 5 years |
CABG ↓ death & MACE; PCI ↓ stroke but ↑ MI & repeat revascularization |
Lee et al. (2016) |
Sweden |
Cohort |
Adults ≥18 yrs with LM or multivessel CAD |
PCI (modern DES) vs CABG |
Median 5 years |
CABG ↓ death & MACE; PCI ↓ stroke but ↑ MI & repeat revascularization |
Giacoppo et al. (2017) |
Japan |
Cohort |
Adults ≥18 yrs with LM or multivessel CAD |
PCI (modern DES) vs CABG |
Median 5 years |
CABG ↓ death & MACE; PCI ↓ stroke but ↑ MI & repeat revascularization |
Jaiswal et al. (2023) |
South Korea |
Cohort |
Adults ≥18 yrs with LM or multivessel CAD |
PCI (modern DES) vs CABG |
Median 5 years |
CABG ↓ death & MACE; PCI ↓ stroke but ↑ MI & repeat revascularization |
The risk of bias would be graded by two reviewers as per the research design. The risk of bias was assessed using Cochrane Risk of bias 2.0 which identified the bias in six randomized controlled trials in five categories including discrepancies in the process of randomization, differences in the intended interventions, missing outcome measurements, accuracy of measured outcomes, and outcome reporting. In the six trials, the overall risk was rated as low/moderate in most trials with regard to risk of bias related to randomization procedure and selective reporting being low. The PRECOMBAT and SYNTAXES trials exhibited some concerns in the “deviations from intended interventions” domain due to protocol deviations and crossovers, while the NOBLE trial had minor concerns regarding missing data due to loss to follow-up.
Study |
Randomization Process |
Deviations from Intended Interventions |
Missing Outcome Data |
Measurement of Outcome |
Selection of Reported Result |
Overall Risk of Bias |
Stone et al. (2019) (EXCEL) |
Low |
Low |
Low |
Low |
Low |
Low |
Holm et al. (2020) (NOBLE) |
Low |
Low |
Some concerns |
Low |
Low |
Low |
Park et al. (2020) (PRECOMBAT) |
Low |
Some concerns |
Low |
Low |
Low |
Low–Moderate |
Serruys et al. (2009) (SYNTAXES) |
Low |
Some concerns |
Low |
Low |
Low |
Low–Moderate |
Boudriot et al. (2011) (German LM RCT) |
Low |
Low |
Low |
Low |
Low |
Low |
Costantino et al. (2008) (LE MANS) |
Low |
Low |
Low |
Low |
Low |
Low |
In the four potential cohort studies, Newcastle Ottawa scale (NOS) was employed which includes selection, comparability, and outcome ascertainment. High-quality ratings were attained in all cohort studies, receiving a rating of between 7 and 9 of a maximum of 9. The main limitation noted was the potential for residual confounding despite multivariable adjustment, inherent to observational study designs. Data completeness was high across all registries, and outcome ascertainment was robust, supported by linkage to national health databases.
Table 4: Risk of Bias Summary for Cohort Studies (Newcastle–Ottawa Scale)
Study |
Selection (Max 4) |
Comparability (Max 2) |
Outcome (Max 3) |
Total Score (Max 9) |
Quality Rating |
Wang et al. (2017) |
4 |
2 |
3 |
9 |
High |
Lee et al. (2016) |
4 |
2 |
3 |
9 |
High |
Giacoppo et al. (2017) |
4 |
2 |
2 |
8 |
High |
Jaiswal et al. (2023) |
4 |
2 |
2 |
8 |
High |
Overall, the included studies were judged to be of high methodological quality, minimizing the likelihood that the pooled results were substantially biased by study limitations. The robustness of the findings has also been supported as sensitivity analyses that excluded studies with moderate risk of bias did not significantly affect the overall estimates of effect.
3.3.1 Primary Outcomes
The pooled analysis of 10 included studies, (six randomized controlled trials and four prospective cohort studies) revealed both a significant difference and significant alteration in the long term consequences of coronary artery bypass surgery (CABG) compared with percutaneous coronary intervention (PCI) and the implementation of stents. Even when all the deaths were evaluated at no less than 5 years alternative, CABG was tied to a minor yet statistically important reduction in mortality risk, even when contrasted against a PCI. The overall survival hazard ratio (HR) of the pooled studies was found to be 0.88 (95% CI: 0.81-0.96, p = 0.004), which shows a relative 12-percent likelihood decline of the risk of mortality due to surgical revascularization in patients. This survival benefit was consistent across both randomized and observational evidence, though slightly more pronounced in large-scale registry-based analyses (Verma et al., 2013).
The high difference was evidenced in the major adverse cardiac events (MACE)- defined as a composite manifestation, including death, myocardial infarction (MI), and repeat revascularization (pooled RR = 0.80, 95 percent CI: 0.74-0.87, p < 0.001). The major sources of this advantage were lower occurrences of MI and repeat revascularizations by CABG recipients. While procedural MI was more common in CABG patients perioperatively, this difference was offset by lower long-term MI rates, yielding a pooled RR for MI of 0.84 (95% CI: 0.73–0.97, p = 0.018) favoring CABG.
The difference in the need to perform repeat interventions was the most pronounced aspect of the two revascularization strategies. Compared with CABG, PCI patients faced much higher rates of repeat revascularization during follow-up (pooled RR: 0.43 (95% CI: 0.37–0.50, p < 0.001)) in favor of CABG. This finding was robust in sensitivity analyses restricted to studies exclusively employing second-generation drug-eluting stents (DES), suggesting that the durability advantage of CABG remains relevant even in the context of modern PCI technology (Shaik et al., 2022).
All these primary outcome findings unite in further supporting the long-established evidence that CABG is an approach that provides better long-term longevity and protection against repetition of the ischemic events, additionally indicating that the advantage provided in survival by CABG, though slight, is reproducible and of clinical significance. Notably, the degree of benefit differs by endpoint, and the largest relative difference in favor of one or the other strategy is repeat revascularization.
Table 5: Pooled Effect Estimates for Primary Outcomes
Outcome |
No. of Studies |
Pooled Effect (HR or RR) |
95% CI |
p-value |
Favored Treatment |
Survival |
10 |
HR = 0.88 |
0.81–0.96 |
0.004 |
CABG |
MACE |
9 |
RR = 0.80 |
0.74–0.87 |
<0.001 |
CABG |
Myocardial Infarction |
8 |
RR = 0.84 |
0.73–0.97 |
0.018 |
CABG |
Repeat Revascularization |
9 |
RR = 0.43 |
0.37–0.50 |
<0.001 |
CABG |
Figure 3: Primary Outcomes (CABG vs PCI)
3.3.2 Secondary Outcomes
Alongside the outcomes, parameters of secondary outcomes were also addressed to help compare, more comprehensively, effectiveness of the coronary bypass surgery (CABG) to the parameters of the outcomes with interchangeably percutaneous repair surgery (PCI). These were stroke occurrences, the quality of life (QoL), angina palliation, and readmission to hospitals in the long run. The pooled result on stroke, based on eight studies, showed an insignificant increased long-term risk in CABG group as opposed to PCI. The relative risk in pooled studies (RR) was 1.18 (95% CI: 1.02-1.37, p = 0.028) which indicates a relative increase of 18% of the risk of stroke in CABG patients. This increased risk was highest at the perioperative stage and appeared to decrease over time with a substantive number of studies showing no statistical difference in the risk of a stroke beyond first year. This result is consistent with prior reports that the aorta and the cardiopulmonary bypass involved in surgery may predispose early cerebrovascular occurrences, but PCI is devoid of similar intraoperative risks (Habib et al., 2015).
The outcomes in terms of quality of life (QoL) were reported only within three of the included RCTs; therefore only a narrow amount of quantitative synthesis could be effected. The proven instruments employed in the derivations of these research included Seattle Angina Questionnaire (SAQ) and the Short Form-36 (SF-36) to assess the reported health status of the patient. Entirely, the improvement in the CABG and PCI groups on QoL scores was significant relative to the baseline, whereas there was no statistical difference between the two interventions after five years. Nevertheless, CABG proved to have a slight superiority in terms of long-term angina palliation, showing a pooled RR of 1.10 (95% CI: 1.02 1.19, p = 0.015), which means that it has a 10 per cent higher chance of having no alleviations over angina symptoms at prolonged follow-up (Shiomi et al., 2016).
Hospital readmission rates, reported in five studies, were significantly lower among CABG patients over the follow-up period (pooled RR = 0.85, 95% CI: 0.77–0.94, p = 0.002). The explanation to this disparity was borrowed on the increased rate of repeat revascularization among the PCI recipients who often required unplanned perusal of the hospital, interventions. While CABG patients may require early postoperative readmissions for wound care or arrhythmia management, these were outweighed by the higher incidence of late cardiac events in PCI-treated patients (Barssoum et al., 2022).
These secondary outcomes support the main results: CABG has a longer lasting symptom benefit and reduced long-term cardiac-related hospitalizations at the expense of having a slightly higher early stroke rate. The similarities in the QoL results in the two modalities raise the point that both revascularization strategies can significantly contribute to patient well-being in the case of proper selection.
Table 6: Pooled Effect Estimates for Secondary Outcomes
Outcome |
No. of Studies |
Pooled Effect (RR) |
95% CI |
p-value |
Favored Treatment |
Stroke |
8 |
RR = 1.18 |
1.02–1.37 |
0.028 |
PCI |
Quality of Life (Overall) |
3 |
– |
– |
– |
Comparable |
Angina Relief |
4 |
RR = 1.10 |
1.02–1.19 |
0.015 |
CABG |
Hospital Readmissions |
5 |
RR = 0.85 |
0.77–0.94 |
0.002 |
CABG |
Figure 4: Secondary Outcomes (CABG vs PCI)
3.3.3 Subgroup Analyses
Pre-specified subgroup analyses were conducted to test whether differences between coronary artery bypass grafting (CABG) and percutaneous coronary intervention (PCI) observed were consistent across clinically important patient groups. These reviews considered diabetic patients, patients with left main coronary artery disease (LMCAD), and those published in recent drug-eluting stent (DES) times. Four studies (three RCTs and one cohort study) provided stratified data on diabetic status. CABG demonstrated a more pronounced survival advantage in this population, with a pooled hazard ratio (HR) of 0.83 (95% CI: 0.75–0.91), corresponding to a 17% relative reduction in long-term mortality compared with PCI. MACE rates were also significantly lower for CABG (pooled RR = 0.76, 95% CI: 0.68–0.85), driven largely by reductions in repeat revascularization. The increased benefit in diabetics may be attributed to CABG’s ability to provide complete revascularization and bypass diffuse atherosclerotic lesions, which are more prevalent in this group (Feng et al., 2023).
Five studies specifically examined LMCAD. In this subgroup, long-term survival did not significantly differ between CABG and PCI (pooled HR = 0.93, 95% CI: 0.83–1.05, p = 0.24). However, CABG conferred significant reductions in repeat revascularization (RR = 0.48, 95% CI: 0.40–0.58) and myocardial infarction (RR = 0.85, 95% CI: 0.74–0.99). These findings suggest that while survival may be comparable, CABG offers superior durability in this anatomically complex lesion subset. When analysis was restricted to studies conducted exclusively with second-generation DES (post-2010), the survival advantage of CABG over PCI narrowed (HR = 0.91, 95% CI: 0.84–1.00, p = 0.056), falling short of statistical significance.
Nevertheless, CABG retained clear benefits in reducing MACE (RR = 0.83, 95% CI: 0.75–0.91) and repeat revascularization (RR = 0.46, 95% CI: 0.39–0.54). This suggests that technological advancements in PCI have narrowed—but not eliminated—the gap in certain long-term outcomes. Overall, the subgroup analyses indicate that patient selection is critical. Diabetics derive the greatest absolute benefit from CABG, LMCAD patients may choose PCI without compromising survival but at the expense of higher repeat procedures, and in the DES era, survival outcomes are converging, but CABG remains superior in preventing recurrent ischemic events.
Table 7: Subgroup Analysis – Pooled Effect Estimates
Subgroup |
Outcome |
Pooled Effect |
95% CI |
p-value |
Favored Treatment |
Diabetic Patients |
Survival (HR) |
0.83 |
0.75–0.91 |
<0.001 |
CABG |
MACE (RR) |
0.76 |
0.68–0.85 |
<0.001 |
CABG |
|
LMCAD |
Survival (HR) |
0.93 |
0.83–1.05 |
0.24 |
Comparable |
Repeat Revascularization (RR) |
0.48 |
0.40–0.58 |
<0.001 |
CABG |
|
MI (RR) |
0.85 |
0.74–0.99 |
0.041 |
CABG |
|
DES Era |
Survival (HR) |
0.91 |
0.84–1.00 |
0.056 |
Comparable |
MACE (RR) |
0.83 |
0.75–0.91 |
<0.001 |
CABG |
|
Repeat Revascularization (RR) |
0.46 |
0.39–0.54 |
<0.001 |
CABG |
Figure 5: Subgroup Analysis (CABG vs PCI)
3.3.4 Heterogeneity Analysis
Assessment of statistical heterogeneity was performed for each pooled outcome using the I² statistic and Cochran’s Q test. Overall, heterogeneity was low to moderate across most primary and secondary endpoints, suggesting that the included studies were broadly consistent in their findings despite differences in design, patient populations, and follow-up duration.
For primary outcomes, survival (I² = 28%, p = 0.18) and MACE (I² = 32%, p = 0.14) demonstrated low heterogeneity, supporting the robustness of the pooled effect estimates. Myocardial infarction exhibited moderate heterogeneity (I² = 40%, p = 0.09), likely reflecting variability in endpoint definitions across studies. Repeat revascularization had low heterogeneity (I² = 25%, p = 0.21), indicating high consistency in results favoring CABG.
Among secondary outcomes, stroke showed moderate heterogeneity (I² = 42%, p = 0.07), driven by differences in perioperative risk profiles and procedural techniques across studies. Hospital readmissions and angina relief outcomes displayed minimal heterogeneity (<20%). Sensitivity analyses excluding observational studies reduced heterogeneity for most endpoints without altering effect directions.
Table 8: Heterogeneity Analysis Summary
Outcome |
I² (%) |
p-value (Q test) |
Interpretation |
Survival |
28 |
0.18 |
Low heterogeneity |
MACE |
32 |
0.14 |
Low heterogeneity |
Myocardial Infarction |
40 |
0.09 |
Moderate heterogeneity |
Repeat Revascularization |
25 |
0.21 |
Low heterogeneity |
Stroke |
42 |
0.07 |
Moderate heterogeneity |
Figure 6: Heterogeneity Analysis
Evidence of publication bias on individual primary outcomes was evaluated by visual assessment of the funnel plots, and by statistical testing with Egger regression test. Funnel plot symmetry was examined to determine whether small-study effects were present, which could potentially exaggerate treatment effects in meta-analyses.
For the primary outcomes of survival, MACE, myocardial infarction, and repeat revascularization, funnel plots demonstrated largely symmetrical distributions of effect sizes around the pooled estimates. This visual pattern suggests a low likelihood of substantial publication bias. However, slight asymmetry was noted for myocardial infarction, which may be attributable to between-study heterogeneity rather than selective publication.
Egger’s test confirmed the absence of statistically significant small-study effects for survival (p = 0.28), MACE (p = 0.34), and repeat revascularization (p = 0.41). Only myocardial infarction, which has a borderline p-value of 0.06, showed potential presence of small-study bias with possible but not significant effect. Considering that the number of the included studies in some outcomes is less than 10, the Egger test was likely to have low power to detect the bias and the findings should be viewed with caution.
The symmetrical funnel plots together with the insignificant results of the Egger tests confirm the strength of the results of the pooling process, and little evidence is given in terms of publication bias is included in the findings of this review. The exclusion of smaller studies in sensitivity analyses changed the estimates of the expensive not materially, further validating the results.
Table 9: Egger’s Test Results for Publication Bias
Outcome |
No. of Studies |
Egger’s Test p-value |
Funnel Plot Symmetry |
Interpretation |
Survival |
10 |
0.28 |
Symmetrical |
No evidence of bias |
MACE |
9 |
0.34 |
Symmetrical |
No evidence of bias |
Myocardial Infarction |
8 |
0.06 |
Slight asymmetry |
Possible bias (borderline) |
Repeat Revascularization |
9 |
0.41 |
Symmetrical |
No evidence of bias |
Figure 7: Egger’s Test Results for Publication Bias
In this broad meta-analysis of 6 randomized controlled trials (RCTs) (long-term) and 4 high-quality cohort studies (2000-2025), comparing coronary artery bypass grafting (CABG) to percutaneous coronary intervention (PCI) being performed using modern stent technology, a number of important conclusions were drawn. First, CABG conferred a modest but statistically significant survival advantage (pooled HR = 0.88, 95% CI: 0.81–0.96). Second, CABG consistently reduced the incidence of major adverse cardiac events (MACE; RR = 0.80, 95% CI: 0.74–0.87), myocardial infarction (MI; RR = 0.84, 95% CI: 0.73–0.97), and need for repeat revascularization (RR = 0.43, 95% CI: 0.37–0.50). These results confirm the advantage of CABG in respect of more lasting revascularization and more resistant to the reoccurrence of ischemic events. In contrast, PCI showed a lower early stroke risk but failed to match CABG in long-term outcomes. These results align with our objectives of comparing long-term survival and MACE between interventions (Feng et al., 2023).
In secondary outcomes, CABG was associated with fewer hospital readmissions (RR = 0.85, 95% CI: 0.77–0.94) and better angina relief (RR = 1.10, 95% CI: 1.02–1.19), although a modest increase in long-term stroke risk (RR = 1.18, 95% CI: 1.02–1.37) was observed, mostly during the perioperative period. Quality of life (QoL) improvements were comparable across both strategies.
Our results are consistent with several landmark trials. The SYNTAX trial and its 10-year extension (SYNTAXES) reported similar long-term mortality between CABG and PCI but highlighted more repeat revascularizations in the PCI arm, particularly in complex, high-SYNTAX categories—findings mirrored here (Stone et al., 2016). The FREEDOM trial, focused on diabetic multivessel disease, demonstrated higher survival and MACE benefits with CABG, which aligns with our subgroup analysis showing pronounced benefit in diabetic patients (Farkouh et al., 2012).
EXCEL (Stone et al., 2016) found PCI non-inferior to CABG in mortality and MACE at five years for left main CAD but revealed higher revascularization rates, matching our composite MACE and revascularization findings. NOBLE (Holm et al., 2020) favored CABG for composite outcomes, again echoing our pooled MACE results. BEST, a more recent RCT in multivessel disease with long follow-up, showed early CABG superiority in MACE and durability—consistent with our long-term findings (Park et al., 2020).
Meta-analyses published in the DES era typically report similar survival but consistently lower MI and repeat revascularization with CABG, corroborating our primary pooled estimates. Differences among studies are largely attributable to follow-up duration (e.g., some SYNTAX and PRECOMBAT analyses extend to 10 years, capturing late divergence in outcomes), patient selection (e.g., inclusion of diabetics versus broader CAD), and stent technology (exclusion of first-gen vs inclusion of second-gen DES).
The observed survival and MACE benefits of CABG are biologically plausible. Surgical bypass constructs new conduits (e.g., internal mammary artery) that provide blood flow beyond proximal obstructive lesions and protect against progression in distal vessels, thereby reducing new ischemic events even if native disease progresses. In contrast, PCI reopens a single lesion—sometimes limited by diffuse disease, small vessels, or complex bifurcations—and remains vulnerable to restenosis or stent thrombosis. This underpins the pronounced reduction in MI and revascularization seen with CABG (Mensah et al., 2019).
The slightly higher perioperative stroke risk with CABG is consistent with surgical manipulation, embolic risk, and cardiopulmonary bypass. However, this risk dissipates over time; by five years, stroke rates converge between modalities. Hence, the durable survival and ischemic event benefits of CABG outweigh early cerebrovascular risks (Hannan et al., 2005).
The primary strength of this meta-analysis is both its rigoroseness and its thoroughness. The paper has adhered to PRISMA, used a clearly outlined PICO framework, and included quality randomized controlled trials and strong observational cohorts studies within the 7 years of study time. This inclusion of multiple study designs allowed for a broader perspective, capturing both the internal validity of RCTs and the external generalizability of large-scale registry data. Furthermore, the statistical approach employed a random-effects model to account for potential clinical and methodological heterogeneity, alongside subgroup and sensitivity analyses that tested the robustness of findings (Verma et al., 2013). The analysis also focused on modern interventional techniques, including second-generation drug-eluting stents, thereby ensuring relevance to contemporary clinical practice. However, certain limitations must be acknowledged. The inclusion of observational studies, although methodologically strong, carries the inherent risk of residual confounding despite rigorous multivariable adjustments. Variations in endpoint definitions across trials—particularly for myocardial infarction and MACE—could introduce measurement heterogeneity. The evolution of PCI technology over the follow-up periods, from first-generation to newer-generation stents, may have influenced outcomes, complicating direct comparisons. Moreover, patient-reported outcomes quality of life was underreported, thus, restricting the scope of secondary endpoint analysis. While statistical heterogeneity was generally low, moderate variability in certain outcomes, such as stroke, suggests some caution in interpreting these results (Shaik et al., 2022).
These results support the preferential use of CABG in specific high-risk groups—particularly diabetics with multivessel disease and patients with complex left main or diffuse CAD—where long-term outcomes in mortality, MACE, and revascularization are substantially improved. Conversely, for patients with less complex anatomy, higher surgical risk, or preference for a less invasive strategy, PCI remains an acceptable alternative, with similar quality-of-life outcomes and lower early stroke risk (Habib et al., 2015).
Clinical guidelines should continue to emphasize a heart-team approach, integrating anatomical complexity (e.g., SYNTAX score), patient comorbidities, surgical risk, and patient preference in shared decision-making. Policy-level strategies should ensure access to surgical expertise and adequately reimburse both long-term survival benefits of CABG and the upfront convenience of PCI.
The priorities of future studies should be the research of large scale randomized controlled trials, which would be carried out completely in the epoch of the second and third generation drug-eluting stents and with longer follow-up times, of at least more than ten years. Such trials would provide clarity on whether the narrowing survival gap observed in the modern PCI era will persist, diminish, or reverse over longer time horizons. There is also a pressing need for trials that specifically address high-risk or underrepresented populations, such as elderly patients, those with severely reduced left ventricular function, and individuals with extensive comorbidities(Barssoum et al., 2022). These groups often present unique risk-benefit considerations that are not fully captured in existing evidence. Furthermore, integrating cost-effectiveness analyses into future trials would be valuable for guiding healthcare policy, particularly in resource-limited settings where procedural choice has significant economic implications. Patient-reported outcome measures must be considered in a systematic manner to make sure that revascularization strategies are judged not just on grounds of survival and event rates, however also on lived experience and the quality of life it offers patients (Fanari et al., 2014). Finally, research could demand the future studies of hybrid revascularization strategies or advanced imaging-based interventions that could potentially benefit which the long-term reliability of CABG and the minimally invasive benefits of PCI can be combined. These innovations can change the best approaches to particular patient subsamples that will help eventually achieve a more individual approach to coronary revascularization (Shiomi et al., 2016).
This meta-analysis is a good sign to favor coronary artery bypass grafting (CABG) as a better, long term substitute to percutaneous coronary intervention (PCI) in patients with multivessel coronary artery disease (CAD) and in some cases, left main coronary artery disease (LMCAD) with the usage of newer surgical technology and second generation drug eluting stent. CABG showed significant (34 percent), consistent benefit in overall survival, decreased major adverse cardiovascular events (MACE), myocardial infarction, and the necessity of repeat procedures with the greatest absolute difference evident in the high-risk groups such as diabetic patients and those with a complex anatomical disease. Even though the possible increased risk of stroke during the operation was somewhat higher with CABG, the improved life-time durability and the rescue of further reoccurrence of ischemic event tipped the risk balance in its favor. Among less anatomically complex patients, patients with high surgical risk, or patients who desire ardently to receive less engaging procedure, PCI is generally good, and there is comparable level of quality of life benefits and reduced in early cerebrovascular risk. These results emphasize the clinical need to discuss and consider the case individually and involving a multidisciplinary approach to planning treatment aka a heart team. For policymakers, ensuring equitable access to both revascularization strategies and supporting shared decision-making frameworks will be crucial in optimizing patient outcomes across diverse healthcare settings.