The optimal timing of coronary artery bypass grafting (CABG) in patients after an acute myocardial infarction (MI) is unknown. We performed a systematic review and meta-analysis of studies comparing mortality rates in patients who underwent CABG at different time intervals after acute MI. Bias assessments were completed for each study, and summary of proportions of all-cause mortality were calculated based on CABG at various time intervals after MI. A total of 22 retrospective studies, which included a total of 137,373 patients were identified. The average proportion of patients who died when CABG was performed within 6 hours of MI was 12.7%, within 6 to 24 hours of MI was 10.9%, within 1 day of MI was 9.8%, any time after 1 day of MI was 3.0%, within 7 days of MI was 5.9%, and any time after 7 days of MI was 2.7%. Interstudy heterogeneity, assessed using I 2 values, showed significant heterogeneity in death rates within subgroups. Only 1 study accounted for immortal time bias, and there was a serious risk of selection bias in all other studies. Confounding was found to be a serious risk for bias in 55% of studies because of a lack of accounting for type of MI, MI severity, or other verified cardiac risk factors. The current publications comparing timing of CABG after MI is at serious risk of bias because of patient selection and confounding, with heterogeneity in both study populations and intervention time intervals.
In patients who develop acute myocardial infarction (MI) with multivessel or left main coronary artery disease, coronary artery bypass grafting (CABG) remains the mainstay of therapy. The current American College of Cardiology/American Heart Association guidelines outline indications for emergent CABG, however, the guidelines do not address optimal timing of CABG in more clinically stable patients after acute MI. Retrospective studies generally show that mortality is lower in patients who receive CABG at later compared with earlier post-MI time points, although the delay can be quite variable depending on the study. In the absence of randomized controlled trials, best practices for optimal timing of CABG after acute MI will continue to be based on retrospective data and expert opinion. The objective of the present study is to conduct a systematic review and meta-analysis of retrospective studies evaluating mortality in relation to the timing of CABG after acute MI.
Methods
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement was followed and is detailed in Supplementary Figure 1 . Studies published between 2000 and 2020 were identified using MEDLINE, the search syntax used for literature review is listed in Supplementary Table 1 . This systematic review was not registered on PROSPERO.
Studies included in the systematic review were nonrandomized retrospective studies that compared mortality in patient groups who underwent CABG after acute MI at varying time points. This included both non-ST-elevation MI (NSTEMI) and ST-elevation MI (STEMI), as some studies included 1 or both clinical scenarios. Only studies that evaluated 30-day or in-hospital mortality were included in the review. Studies were excluded if they were systematic reviews or meta-analyses, incorporated percutaneous coronary intervention as an intervention, or included an additional concomitant surgery in setting of CABG (e.g., valvular repair).
A bias assessment was completed for each of the included studies and evaluated 7 domains. Details regarding the domains used for bias assessment are listed in Table 1 . The Risk of Bias in Nonrandomized Studies of Intervention assessment tool employs signaling questions to evaluate for potential bias in each domain, and judgments carry forward to form an overall evaluation regarding the risk of bias. With the available information, each domain can be determined to be at low, moderate, serious, or critical risk for bias; and a designation of “no information” can be used if a judgment cannot be formed. Each study was evaluated independently by 2 reviewers (FW, AF) who adjudicated risk of bias in each domain. Disputes were resolved by a third reviewer (MN).
| ROBINS-I bias assessment for timing of CABG following acute MI | |
|---|---|
| Bias domain | Criteria reviewed |
| Confounding | MI severity (inotropic/vasopressor support, Killip classification, etc.) MI type (STEMI vs NSTEMI) Verified cardiovascular risk factors (LVEF, renal disease, COPD, liver disease, HTN, DM, age, gender) |
| Selection | Immortal time bias |
| Classification of Intervention | Time-interval definitions |
| Deviations from Intended Intervention | Alternative procedure (PCI, Valve repair, LVAD, Impella, etc.) Change in timing of intervention based on clinical status |
| Missing Data | Loss of follow-up |
| Measurement of Outcomes | Measurement technique for outcomes (30-day mortality, hospital mortality) |
| Selection of Reported Results | Reporting results from subgroups Reporting results after multiple sub-analyses |
As the studies included in this review employed markedly different time intervals as comparators, the decision was made to forego traditional meta-analysis. Instead, a proportion meta-analysis was utilized to assess mortality in time-based subgroups to provide quantitative data to improve risk-assessment and surgical prognostication using the available information. Fixed-effects and random-effects proportion meta-analysis was performed using the Freeman-Tukey transformation and weighted summary proportions and 95% confidence intervals (CIs) were calculated. This method transforms the data into new variables to form a normal distribution and also stabilizes the variance of the population.
The transformed data are compared with percentages of the normal distribution to report the 95% CI. The proportion of outcome variance between studies attributable to interstudy heterogeneity was assessed by calculating I 2 values. The outcome analyzed in the meta-analysis was all-cause mortality. The following subgroups were analyzed: CABG within 6 hours of MI, within 6 to 24 hours of MI, any time within 1 day of MI, any time after 1 day of MI, any time within 7 days of MI, and any time after 7 days of MI. Studies were also evaluated in subgroups for publication bias with funnel plots. This meta-analysis was performed using MedCalc for Windows, version 15.0 (MedCalc Software, Ostend, Belgium).
Results
In total 22 retrospective studies, evaluating 137,373 patients, were identified evaluating either 30-day or in-hospital mortality in patients who underwent CABG after acute MI. A total of 9 studies were multi-center registries or population-level databases, whereas 13 studies were from a single institution. A total of 12 studies evaluated CABG after MI (nonspecified), whereas 4 studied CABG after STEMI. CABG after NSTEMI was studied in 5 reports, and CABG after transmural MI was evaluated in 1 report.
The general characteristics of the studies included in this review are outlined in Table 2 . A total of 10 studies reported a statistically significant mortality benefit in the delayed CABG groups, although the intervention groups (time intervals from MI to CABG) were not consistently defined throughout these studies. There was no difference in the overall risk of bias for studies that reported a mortality benefit with delayed CABG compared with those studies that did not report a mortality difference with delayed CABG. Eight studies that reported a statistically significant mortality benefit evaluated CABG after MI (nonspecific), 2 studies evaluated CABG after STEMI.
| Study | Design | Experimental setup | Results | ||||
|---|---|---|---|---|---|---|---|
| First Author | Year | Study Type | Participants | Intervention | Outcome | Comparator | Conclusion |
| Lee | 2001 | Observational | 44,365 (NY State Cardiac Surgery Registry) | CABG AFTER MI | Hospital Mortality | < 6h (11.8%), 6-23h (9.5%), > 1d (2.8%) | Improved Survival with Delay of CABG |
| Lee | 2003 | Observational | 32,099 (NY State Cardiac Surgery Registry) | CABG AFTER MI (TRANSMURAL MI) | Hospital Mortality | < 6h (14.2%), 6-23h (13.8%), 1-3d (7.9%), 4-7d (3.8%), 8-14d (2.9%), > 15d (2.7%) | Improved Survival with Delay of CABG |
| Voisine | 2005 | Observational | 13,545 (Single Institution) | CABG AFTER MI | 30-Day Mortality | < 6h (19.2%), 6-24h (9.8%), 1-7d (8.6%), 8-30d (3.2%), > 30d (2.4%) | Improved Survival with Delay of CABG |
| Kamohara | 2006 | Observational | 67 (Single Institution) | CABG AFTER MI | 30-Day Mortality | < 6h (9.1%), 6-24h (2.9%) | No Mortality Difference |
| Chew | 2007 | Observational | 9,053 (SYNERGY Trial) | CABG AFTER MI (NSTEMI) | 30-Day Morality | < 72h (4.6%), ≥ 72h (3.6%) | No Mortality Difference |
| Thielmann | 2007 | Observational | 138 (Single Institution) | CABG AFTER MI (STEMI) | Hospital Mortality | ≤ 6h (10.8%), 7-24h (23.8%), 1-3d (6.7%), 4-7d (4.2%), 8-14d (2.4%) | Improved Survival with Delay of CABG |
| Weiss | 2008 | Observational | 9,476 (California Discharge Data) | CABG AFTER MI | Hospital Mortality | 0-2d (5.6%), ≥ 3d (3.8%) | Improved Survival with Delay of CABG |
| Abd-Alaal | 2010 | Observational | 278 (Single Institution) | CABG AFTER MI | 30-Day Mortality | ≤ 24h (11.7%), 24-72h (7%), > 2w (2.5%) | Improved Survival with Delay of CABG |
| Parikh | 2010 | Observational | 2,647 (ACTION Registry-GWTG) | CABG AFTER MI (NSTEMI) | Hospital Mortality | ≤ 48h (3.6%), > 48h (3.8%) | No Mortality Difference |
| Assmann | 2012 | Observational | 1,168 (Single Institution) | CABG AFTER MI | 30-Day Mortality | < 6h (14.8%), 6-24h (10.2%), 2-3d (8.8%), 4-10d (4.2%), 11-20d (2.3%), 21-30d (2.0%) | Improved Survival with Delay of CABG |
| Ngaage | 2013 | Observational | 8320 (Single Institution) | CABG AFTER MI | Hospital Mortality | 0-30d (3.5%), 31-90d (2.6%), > 90d (1.2%), no MI (1.1%) | Improved Survival with Delay of CABG |
| Clark | 2015 | Observational | 120 (Single Institution) | CABG AFTER MI | Hospital Mortality | ≤ 5d (6.7%), > 5d (1.3%) | No Mortality Difference |
| Davierwala | 2015 | Observational | 758 (Single Institution) | CABG AFTER MI (NSTEMI) | Hospital Mortality | < 24h (6.0%), 24-72h (4.7%), 3-21d (5.1%), | No Mortality Difference |
| Dudek | 2015 | Observational | 314 (ACCOAST Trial) | CABG AFTER MI (NSTEMI) | 30-Day Morality | < 2.98d (3.9%), 2.98-6.95d (1.9%), ≥ 6.95d (1.9%) | No Mortality Difference |
| Khan | 2015 | Observational | 184 (Single Institution) | CABG AFTER MI (STEMI) | 1-Month Mortality | ≤ 24h (18.1%), > 24h (11.4%) | No Mortality Difference |
| Ha | 2017 | Observational | 3,110 (National Inpatient Sample) | CAGB AFTER MI (NSTEMI) | Hospital Mortality | ≤ 48h (2.0%), > 48h (1.8%) | No Mortality Difference |
| Nichols | 2017 | Observational | 3,060 (NNE Cardiac Surgery Registry) | CABG AFTER MI | Hospital Mortality | < 1d (5.1%), 1-2d (1.6%), 3-7d (1.6%), 8-21d (2.7%) | Improved Survival with Delay of CABG |
| Rohn | 2017 | Observational | 135 (Single Institution) | CABG AFTER MI (STEMI) | 30-Day Mortality | ≤ 6h (5.7%), 6-24h (9%) | No Mortality Difference |
| Wang | 2017 | Observational | 264 (Single Institution) | CABG AFTER MI (STEMI) | 30-Day Mortality | < 21d (6.5%), 21-90d (0%), > 90d (2.4%) | No Mortality Difference |
| Bianco | 2019 | Observational | 2058 (Single Institution) | CABG AFTER MI | 30-Day Mortality | < 24h (4.17%), > 24h (4.58%) | No Mortality Difference |
| Al-Omary | 2020 | Observational | 251 (Single Institution) | CABG AFTER MI | 30-Day Mortality | ≤ 7d (1.43%), > 7d (4.76%) | No Mortality Difference |
| Lemaire | 2020 | Observational | 5,963 (National Inpatient Sample) | CABG AFTER MI (STEMI) | Hospital Mortality | < 1d (8.2%), 2-3d (3.53%), 4-7d (2.94%) | Improved Survival with Delay of CABG |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
