Survival and outcomes of coronary artery bypass grafting in patients with extremely low ejection fraction (≤20%): A systematic review and meta-analysis Authors: Wenteng Hou, Iris Liu, Nader S Aboelnazar, Rashmi Nedadur, Thin Xuan Vo, A Dave Nagpal Published: 2026 ABSTRACT AbstractObjectivesTo assess perioperative outcomes and long-term survival of coronary artery bypass in patients with ejection fraction (EF) 20% or less.MethodsWe performed a systematic search of MEDLINE, EMBASE, and CENTRAL via OVID for both randomized controlled trials and cohort studies involving patients with EF ≤20% undergoing coronary artery bypass grafting, including results from 1995 to December 2023. Random-effects prevalence meta-analysis was performed using the meta library in R. Outcome prevalence was logit-transformed and pooled using a generalized linear mixed-effects model. The presence of heterogeneity was assessed using the Cochran Q test with a significance of P < .10 following Cochrane Handbook recommendations and further quantified using I2 statistic.ResultsA total of 29 studies involving 5910 patients were included in this analysis. The pooled prevalence of perioperative mortality was 9%. Postoperative myocardial infarction and stroke prevalence were 2% and 3%, respectively. Low cardiac output syndrome definition varied across studies, and its pooled prevalence was 16%, with postoperative mechanical circulatory support requirement at 9%, the pooled 5-year survival rate was 74%.ConclusionsDespite elevated short-term risk, patients with severely low LVEF (≤20%) may derive long-term survival benefit from coronary artery bypass grafting and therefore it should remain a consideration in advanced heart failure therapies.Key Words: coronary revascularization, heart failure, ischemic cardiomyopathy Five-year survival rate for advanced heart failure therapies. Central Message Despite elevated short-term risk, patients with severely low LVEF (≤20%) may derive long-term survival benefit from CABG, and it should remain a consideration in advanced heart failure therapies. Perspective Patients with LVEF ≤20% are underrepresented in surgical revascularization trials. With good outcomes in transplant and LVAD, more data are required for decision-making in patients with ischemic cardiomyopathy. Our analysis shows that select patients with EF ≤20% can achieve comparable long-term survival with surgical revascularization; therefore, CABG should not be overlooked as a therapeutic option. Heart failure with reduced ejection fraction (EF) is a significant risk factor for increased perioperative morbidity and mortality after cardiac surgery. Abundant data have established improved long-term survival with coronary artery bypass grafting (CABG) combined with medical therapy compared with medical therapy alone in patients with severe coronary artery disease and left ventricular dysfunction.1,2 As such, CABG remains an important therapy for many patients with heart failure with reduced EF despite its short-term risks. Patients with very low EF, defined as EF ≤20%, are a specific group of patients who have been historically underrepresented in studies of surgical revascularization and generally have been only included in larger studies of heart failure with greater EF cutoffs.3 These patients are frequently considered poor surgical candidates, perhaps as a result of their complex medical history, or the low EF itself, given an all-cause mortality risk increasing by 39% for every 10% reduction in EF below ≤45%.4 In addition, patients with EF ≤20% may qualify for alternative advanced heart failure therapies, such as transplantation or durable left ventricular assist device (LVAD).5 These advanced heart failure therapies have shown consistent improvement in survival, with Canadian average 5-year survival for transplant at 85% and durable LVAD at 64% according to the most recent Society of Thoracic Surgeons data, even though there is significant change in quality of life, such as the presence of frequent driveline infections or intense medical surveillance on immunosuppressive therapy.6,7 The lack of trial data focused on the perioperative and long-term outcomes of surgical revascularization in this very low EF subgroup has rendered CABG a controversial and underresearched therapeutic option for this subgroup of patients. The objective of our study was to aggregate all available published data for patients with EF ≤20% undergoing CABG, including both focused study and subgroup data from larger studies, to examine perioperative and long-term outcomes through a systematic review and meta-analysis. Methods This systematic review was conducted according to Cochrane Handbook for Systematic Reviews of Interventions and laid out within Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. This systematic review was prospectively registered on PROSPERO (CRD42024588437). Because this is a systematic review and no patient-level data were used, no institutional review board approval was required. Literature Search We conducted a librarian-assisted comprehensive database search of MEDLINE, EMBASE, and CENTRAL using OVID for all randomized controlled trials (RCTs) and observational cohort studies from inception to December 2023. We hand-searched the bibliography sections of included studies and similar systematic reviews for relevant publications and gray literature. Eligibility Criteria We included both RCTs and observational cohort studies if they met the following criteria: (1) primary procedure was adult (patients ≥18 years old) coronary artery bypass surgery; and (2) at least 1 outcome in patients with EF ≤20%, either as a subgroup or the sole focus of the study. We retroactively excluded all studies published before 1995, given the more heterogeneous surgical practice before this time. No restrictions were placed on language or country of origin. We excluded studies that did not fit the aforementioned criteria: pediatric patients (<18 years old), EF >20%, or patents who received percutaneous coronary intervention. Study Selection After deduplication using Covidence, studies were screened against the eligibility criteria. Publications were independently screened by 2 authors (W.H., I.L.) on the basis of title and abstract. Screened studies were subsequently entered into the full-text screening process before being included in the study. Disagreements during both stages were resolved via discussion involving a third author (N.A.). Data Extraction Data retrieval was completed by 2 independent authors in duplicate (W.H., I.L.) using an a priori designed standardized data extraction form based on Excel (Microsoft). This form was prepiloted using 5 studies from the included full texts. Categories of data collected included study metadata, preoperative data, perioperative data, postoperative outcomes, and long-term outcomes. Risk of Bias Assessment We conducted risk of bias assessment using the newest version of the Risk Of Bias In Non-randomised Studies - of Interventions tool, version 2 (ROBINS-I V2) for observational cohort studies.8 Because we planned to include subgroup data within our analysis, only studies solely focusing on EF ≤20% patients were assessed with the tool, because ROBINS-I V2 was not designed to assess the risk of bias-associated subgroup data. Statistical Analysis We performed random-effects prevalence meta-analysis using the meta library in R.9 Prevalence values were logit-transformed and pooled using a generalized linear mixed-effects model.10 The presence of heterogeneity was assessed using the Cochran Q test with a significance of P < .10 following Cochrane Handbook recommendations.11 Heterogeneity was further quantified using I2 statistics. An I2 value of ≥75% was considered indicated of serious heterogeneity, as recommended by the Cochrane Handbook. The presence of small-study effects as an indication for publication bias was assessed using funnel plots and Egger's regression test.12 The Egger's test was not conducted for meta-analyses with <10 included studies, because the test lacks power to detect real funnel plot asymmetry when a small number of studies are included.13 Results Study Characteristics Our search strategy yielded 4484 studies, which were screened against the eligibility criteria to 34 studies for the data-extraction phase. A total of 29 studies involving 5910 patients were included in the final analysis as the year restriction of 1995 or later was applied after discussions among the author team. The detailed Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart can be seen in Figure E1. Nineteen publications were larger studies that included patients with EF ≤20% as a subgroup with analyzable data,14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32 whereas 10 studies were solely focused on this population.33, 34, 35, 36, 37, 38, 39, 40,E1,E2 Of the full cohort studies included, most compared outcomes of patients with EF ≤20% with those having normal EF, for which we only included the ≤20% group in our analysis. Two publications were RCTs, whereas the rest were observational cohort studies.15,39 The study characteristics can be seen in Table 1. Nine studies was assessed via the ROBINS-I V2 tool, with 6 studies yielding serious risk of bias, 2 with moderate risk of bias, and 1 with critical risk of bias (Table E1). Preferred Reporting Items for Systematic reviews and Meta-Analyses flow chart for the screening process. Table 1Study characteristics for all studies included in our analysisStudy-IDYearCountryStudy typeFocus on ≤20% EFStudy periodIntervention types≤20% EF sample sizeAlgarni 2011142011CanadaRetrospective cohortSubgroupJan 1, 1990, to Dec 31, 2009CABG646Argenziano 1999151999USARCT (CABG Patch trial database)SubgroupMarch 12, 1993, to Feb 5, 1996CABG143Baumgartner 1998161998USAObservationalSubgroup1990-1996CABG31Caputti 2011332011BrazilObservationalFull studyJan 2001 to Dec 2005Off- vs on-pump CABG217Carr 2002342002USAObservationalFull studyJan 1990 to Dec 1999CABG86DiCarli 1998171998USAObservationalSubgroupFeb 1989 to Nov 1991CABG6Dietl 1996181996USAObservationalSubgroupJan 1991 to Dec 1995CABG with or without IABP71Elefteriades 1997191997USAObservationalSubgroupJan 1986 to 1994CABG41Fukunaga 2020352020CanadaObservationalFull study (compared to normal EF)2010-2016CABG143Gatti 2016202016ItalyObservationalSubgroupJan 1, 1990, to Sept 30, 2014CABG23Hovnanian 2010212010BrazilObservationalSubgroup1998-2004CABG31Karimi 2010362010IranObservationalFull study (abstract)Feb 2002 to Jun 2009CABG119Kaul 1996371996USAObservationalFull studyJan 1, 1987, to Dec 31, 1992CABG210Kawajiri 2017382017CanadaObservationalFull studyMay 2002 to April 2012Transplant eligible vs ineligible133Levin 2012392012USARCTFull studyDec 1, 2002, to Jun 1, 2008With or without levosimendan252Mahesh 2016222016UKObservationalSubgroupJuly 1993 to June 2012CABG86Maslow 2002232002USAObservationalSubgroup1994-1998CABG7Mickleborough 1995401995CanadaObservationalFull studyJan 1982 to Mar 1993CABG79Mickleborough 2000E12000CanadaObservationalFull studyJan 1982 to Sept 1997CABG125Nardi 2009242009ItalyObservationalSubgroupJan 1993 to Dec 2003CABG52Pagano 1998251998UKObservationalSubgroupN/ACABG13Shapira 2006262006USAObservationalSubgroup1995-2000CABG81Shum-Tim 1999E21999CanadaObservationalFull studyJan 1991 to Dec 1994CABG14Sternik 1997271997IsraelObservationalSubgroupJan 1991 to July 1994Off- vs on-pump CABG14Tjan 2000282000GermanyObservationalSubgroupJan 1990 to Dec 1998CABG51Topkara 2005322005USAObservationalSubgroup1997-1999CABG2442Vickneson 2019292019SingaporeObservationalSubgroupJan 2009 to Dec 2015CABG114Yau 1999301999CanadaObservationalSubgroup1982-1997CABG680Yoon 2010312010USAObservationalSubgroupN/ACABGN/AEF, Ejection fraction; CABG, coronary artery bypass graft; RCT, randomized control trial; IABP, intra-aortic balloon pump; N/A, not available. Study characteristics for all studies included in our analysis EF, Ejection fraction; CABG, coronary artery bypass graft; RCT, randomized control trial; IABP, intra-aortic balloon pump; N/A, not available. Short-Term Postoperative Outcomes A total of 22 studies involving 4976 patients were included the analysis for postoperative mortality.15,16,19, 20, 21, 22, 23, 24,26, 27, 28, 29, 30,32, 33, 34, 35,37, 38, 39,E1,E2 The pooled prevalence was 9% (95% CI, 0.08-0.11, I2: 45%). The timing of publication ranged from 1996 to 2020. There was no temporal trend in postoperative mortality upon visual inspection, as seen in the forest plot (Figure 1). There was no evidence of publication bias on funnel plot (Egger's P = .3534) (Figure E2). Forest plot for postoperative mortality prevalence in patients with ejection fraction ≤20%. Funnel plot for postoperative mortality. Ten publications ranging from 1999 to 2019 were included in the analysis for postoperative myocardial infarction, involving a total of 4134 patients.22,26,28, 29, 30,32, 33, 34,39,E1 The pooled prevalence was 2% (95% CI, 0.01-0.04, I2: 53%). In contrast, the pooled prevalence of stroke was 3% (95% CI, 0.02-0.03, I2: 0%), aggregated from a total of 10 studies involving 4106 patients from 1998 to 2019.16,21,26,29,30,32, 33, 34,39,E1 There was no evidence of publication bias for both myocardial infarction and stroke on their respective forest plots (Figures E3 and E5) and funnel plots (Figures E4 and E6). Forest plot for postoperative myocardial infarction prevalence. Forest plot for postoperative stroke prevalence. Funnel plot for postoperative myocardial infarction. Funnel plot for postoperative stroke. The definition of low cardiac output syndrome varied between studies, leading to significant heterogeneity. A total of 1362 patients from 6 studies showed a pooled prevalence of low cardiac output syndrome at 16% (95% CI, 0.11-0.22, I2: 84%).22,30,34,38,39,E1 (Figures E7 and E8). The pooled prevalence of patients requiring postoperative mechanical circulatory support was 9% (95% CI, 0.03-0.21, I2: 90%), aggregated from 13 studies involving 4113 patients, with significant heterogeneity.16,21, 22, 23,28,30,32, 33, 34,37,38,E1,E2 (Figures E9 and E10). The definition of postoperative mechanical circulatory support also varied from intra-aortic balloon pump, Impella, to temporary LVAD implantations. Forest plot for postoperative low cardiac output syndrome prevalence. Funnel plot for postoperative low cardiac output syndrome. Forest plot for postoperative mechanical circulatory support requirement prevalence. Funnel plot for postoperative mechanical circulatory support requirement. Long-Term Survival A total of 7 studies involving 529 patients reported 5-year survival data, all of which were included in the analysis, showing a pooled 5-year survival rate of 74% (95% CI, 0.69-0.78, I2: 37%), with mild heterogeneity, as seen in the forest plot within Figure 2.17,24,26,34,36,38,E1 There was no publication bias on visual inspection of the funnel plot (Figure E11), as an Egger's test could not be completed. Only 3 publications reported data regarding 10-year survival. They were Carr and colleagues,34 Kawajiri and colleagues,38 and Mahesh and colleagues,22 reporting 33%, 39%, and 32.3% 10-year survival rates, respectively. Forest plot for 5-year survival in patients with ejection fraction ≤20%. Funnel plot for 5-year survival. Discussion This report represents the most comprehensive systematic review of short- and long-term surgical revascularization outcomes in patients with very low EF (≤20%). Our analysis demonstrates that patients with EF ≤20% undergoing CABG face a perioperative mortality rate of 9%. By comparison, the average mortality for isolated CABG was 1.4% in Canada in 2019, and 2.1% in the United States in 2023.E3 The significantly increased short-term mortality rate in left ventricular ejection fraction ≤20% patients undergoing CABG is unsurprising, given the end-organ dysfunction that often accompanies these patients. Despite the short-term mortality, our analysis indicates that CABG for patients with left ventricular ejection fraction ≤20% may provide a long-term survival benefit, with 5-year survival rate of 74% based on 7 meta-analyzed studies. These results are congruent with the long-term benefits of CABG elucidated in the follow-up analysis of Surgical Treatment for Ischemic Heart Failure (STICH), especially for more advanced forms of ischemic cardiomyopathy, including triple-vessel disease and left ventricular systolic dysfunction.E4,E5 These results are reflected in the 2021 American Heart Association guidelines in coronary revascularization, which maintains a class I recommendation for CABG for patients with severe systolic dysfunction (≤35%) and multivessel disease.E6 An area for investigation is whether the short-term morbidity and mortality can be reduced with more liberal use of prophylactic temporary mechanical circulatory support, as evidenced by the European Association for Cardio-Thoracic Surgery Expert Consensus on Protected Cardiac Surgery document and ongoing investigations such as the IMPACT trial.E7,E8 The 5-year survival rate after a heart transplant is 72% based on international data in several studies,E9,E10 and the most recent Society of Thoracic Surgeons registry data for magnetically levitated LVADs showed a 5-year survival rate of 64%.7 It is crucial to acknowledge that these survival observations cannot be directly compared against our meta-analysis results, because they are composed of intrinsically different patient populations. However, one might cautiously infer that in select patients with EF ≤20%, CABG may have comparable long-term survival versus those who received LVAD or heart transplant. A potential comparison can be made between the survival of patients with specifically ischemic cardiomyopathy who received LVADs or transplants and our CABG cohort, although it would still remain an imperfect comparison. The benefit of revascularization for patients with heart failure was originally understood as an improvement in blood flow to the “hibernating” myocardium, especially in patients with ischemic cardiomyopathy.E11 However, studies such as STICH have questioned the value of determining myocardial viability preoperatively, and that targeted revascularization using myocardial viability may yield similar results as those who did not undergo viability testing.E12 The Revascularization for Ischemic Ventricular Dysfunction (REVIVED) trial revealed that percutaneous coronary intervention plus optimal medical therapy had no benefit over optimal medical therapy alone in patients with EF ≤35% in terms of mortality, myocardial infarction, or repeat revascularizations at a mean of 41 months.E13 Hence, the hibernating myocardium theory cannot fully explain the survival benefit of surgical revascularization in these patients. Instead, prevention of future ischemic events may explain why CABG is associated with significantly lower risks for mortality, MI, and repeat revascularization rates.2,E14,E15 There are several strengths of this review. To increase our search for all available data for patients with EF ≤20% undergoing CABG, we included patients who were originally part of larger cohort studies involving greater EF cut-offs but had analyzable individual group data. For any ≤20% subgroups that were involved in comparative studies, only their individual group outcomes were meta-analyzed. These efforts significantly increased the breadth of our search, as studies solely focusing on EF ≤20% patients were rarer and often older. Our study has several limitations. Conclusions that can be drawn from our single-arm meta-analysis limits our findings to predominantly descriptive reporting and no direct comparisons can be made of CABG against other advanced heart failure therapies. In addition, the significant advancements in heart failure therapies since 1995, such as the addition of angiotensin receptor/neprilysin inhibitor and sodium-glucose cotransporter-2 inhibitor medications, cannot be accounted for in our analysis. In some of our analyses, the results are dominated by a single study (Topkara and colleagues32), accounting for more than 50% of the combined sample size, which would potentially bias the results. In addition, because we included a large amount of subgroup data, we could not assess the risk of bias for all included studies, only those whose sole focus was EF ≤20%. Because only noncomparative individual group data were included in our pooled analysis, we do not foresee any major methodologic pitfalls of this approach. We believe that information gained with the inclusion of these patients in our analyses outweighs potential drawbacks. In conclusion, our systematic review and meta-analysis shows that surgical revascularization for patients with EF ≤20% is associated with a greater perioperative mortality rate but may portend significant long-term survival benefits. Therefore, CABG remains an important option for patients with severely reduced EF. Further prospective investigation focused specifically on this very low-EF patient population is needed to elucidate the short- and long-term outcomes comparing CABG and advanced heart failure therapies to inform individual patient decision-making and future guidelines. Conflict of Interest Statement The authors reported no conflicts of interest. The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest. Appendix E1 Table E1Risk of bias assessment table for studies includedStudy ID:Caputti 201133Carr 200234Fukunaga 202035Karimi 201036Kaul 199637Kawajiri 201738Mickleborough 199540Mickleborough 2000E1Shum-Tim 1999E2Risk of bias due to confounding.SeriousSeriousModerateHIGHModerateModerateSeriousSeriousSeriousRisk of bias in participant selectionModerateLowModerateHighModerateSeriousSeriousSeriousModerateRisk of bias in intervention classificationLowLowLowLowLowLowLowLowModerateRisk of bias due to deviation from intended interventionSeriousModerateLowHighLowLowLowLowLowRisk of bias due to missing dataLowLowLowHighLowLowModerateModerateLowRisk of bias due to outcome measurementLowModerateModerateLowLowLowModerateModerateLowRisk of bias in selective reporting.LowLowLowHighLowLowLowLowLowOverall Risk of Bias:SeriousSeriousModerateCriticalModerateSeriousSeriousSeriousSeriousOnly studies fully focused on patients with EF ≤20% were assessed. Risk of bias assessment table for studies included Only studies fully focused on patients with EF ≤20% were assessed. References E-References