Strengths and limitations of this study

  • Comprehensive sample from 2000 to 2019 included.

  • Only articles with well-defined study groups included.

  • Mechanisms for associations extracted.

  • Scoping review according to PRISMA-ScR guidelines.

Introduction

Coronary artery bypass grafting (CABG) is a safe treatment for patients with coronary artery disease [1]. Progressive improvements in post-operative survival over its fifty-year history have been observed through improvements in surgical technique [2, 3]. Canadian reports estimate 30-day mortality after isolated CABG surgery is 1.3% [4]. However, regional variation in mortality has been observed in Canada [4, 5], the United States [6, 7], and the United Kingdom [8]. Previous research reported numerous factors contributing to this variation. Indeed, fewer deaths are reported in patients with favorable case mix characteristics, including those who are younger [9], have normal ejection fractions [10] and are free of iron deficiency [11]. Care-related factors, including higher hospital and surgeon volume [12, 13], and process factors, such as the use of arterial grafting strategies [14], are also shown to contribute to lower mortality.

However, the entirety of this evidence has not been mapped on the conceptual framework of quality improvement. Without such mapping, designing interventions to improve care quality could be misguided [15].

Many quality improvement initiatives use the Donabedian framework [16] that considers factors related to structures and processes of care. In this framework, structures include the care provider organizational features (services, size, systems, and volume), the human resources (experience and qualifications) and the material resources (equipment, facilities, and staffing ratios) required to provide care. Processes refer to managerial activities (prioritization, scheduling, and discharge planning) and the medical procedures (both diagnostic and treatment) that constitute care delivery within the defined structures. Outcomes refer to the results that may stem from exposure to a factor; in this report we refer to intermediary outcomes to identify factors that occur after exposure to CABG but prior to in-hospital mortality as the terminal outcome. Shroyer et al. [17] augmented this framework by including factors related to patients and disease, which we refer to as ‘case-mix’ in this report.

Scoping reviews use a systematic approach to map evidence on a topic and identify main concepts, as well as knowledge gaps [18]. We use the scoping review methodology to select factors of in-hospital mortality for patients undergoing isolated CABG and map them to the augmented Donabedian framework. We then synthesize information on mechanisms for their effects.

Methods

This review adheres to the Scoping Review extension of the Preferred Reporting Items for Systematic Review and Meta-Analysis statement (PRISMA-ScR) [18].

Eligibility criteria

We included observational studies which reported the association from the regression analysis of postoperative in-hospital mortality among patients aged 19 years and older who underwent isolated CABG, published in English between January 1, 2000 and December 31, 2019 (Table 1).

Table 1 Selection criteria for the literature search
Full size table

We defined risk factors as any attribute, characteristic, or exposure that increases the likelihood of developing a disease or incurring an injury [19]. We excluded intervention studies, any composite outcome of complications and mortality, a study endpoint outside of the hospital setting, and studies where no statistical association was found.

Information sources and search strategy

We searched the electronic databases PubMed, CINAHL, and EMBASE for studies published between 2000 and 2019. Reference lists of retrieved studies were further screened to identify additional studies that may have been missed during database searches.

Search

The search was developed using terms for the intervention (coronary artery bypass graft), outcome (mortality), study design (observational), and analysis (regression) (see Additional file 1 for full electronic search strategy for each database).

Selection of sources of evidence

We exported citations from databases into reference management software for de-duplication prior to screening. Two reviewers independently screened all abstracts against inclusion and exclusion criteria. Conflicts were resolved by consensus. Full texts of potentially eligible studies were independently screened by two reviewers with conflicts resolved by consensus.

Data chart processing and data items

We used a pre-designed form to collect data; the form was piloted by two reviewers on five articles. No conflicts in data extraction were noted. Data extracted included the author’s name, publication date, country, study population, study design, sample size, source of data, risk factor measurement, outcome, and effect estimate. We identified risk factors representing the exposure, treatment, or intervention of primary interest in the title or objectives of the selected papers. We extracted the effect of the primary risk factor from multivariable analysis from papers with well-defined study groups [20]. This was done to avoid misclassification of covariates in multivariable analyses as primary factors [21]. Factors were considered statistically significant when a p-value < 0.05 was reported. The proposed mechanisms for reported associations were extracted from the discussion section by one reviewer. The extraction was checked for accuracy by a second reviewer.

Synthesis of results

We summarized the data in text, tables, and figures. Two authors sorted factors according to common properties, creating 14 unique groups. Two authors then mapped the groups to case-mix, structures, processes, and intermediary outcomes of the augmented Donabedian framework. Disagreements on sorting, grouping, and labelling were resolved by consensus. Finally, we synthesized proposed mechanisms, where mechanisms were reported, for the association between factors and in-hospital mortality from articles in which they were identified (Table 3).

Patient and public involvement

Patients were not involved in the design, conduct, reporting, or dissemination plans of our research.

Results

Search results

The search produced 1773 articles for initial title and abstract screening (Fig. 1). We excluded 1107 articles on title and abstract screening: 582 were not isolated CABG; 525 were not in-hospital mortality. We further excluded 583 articles on full-text screening: 33 had no full-text available, 98 were not isolated CABG, 227 had an outcome that was not in-hospital mortality, 110 used an analysis that was not multivariable regression, and 115 had results where there was no statistical association found. 83 articles remained to be included in the review.

Fig. 1
figure 1

Flow chart of the literature, retrieval, review, exclusion, and selection process

Full size image

Structural factors of in-hospital mortality

In total, 12 articles reported on structural factors of in-hospital mortality after CABG. Factors in these articles were grouped to treatment era [earlier year of operation (n = 4)], care setting [hospital volume (n = 4), hospital type (n = 1)], and operator qualification [operator volume (n = 2), surgeon experience (n = 1)] (Table 2). Of the 6 identified factors, we synthesized 4 mechanisms from 7 articles for their effect on mortality (Table 3).

Table 2 Grouped factors of postoperative mortality in coronary bypass surgery by reviewed article
Full size table
Table 3 Synthesized mechanisms proposed for case-mix characteristics, structures, processes, and intermediary outcomes in reviewed articles
Full size table

Process factors of in-hospital mortality

In total, 27 articles reported on process factors of in-hospital mortality after CABG. Factors in these articles were grouped to pre-operative care [aprotinin (n = 1), ASA (n = 3), beta blockers (n = 1), insulin infusion (n = 1), intra-aortic balloon pump (n = 1), statin (n = 4)], intraoperative management [allogenic blood transfusion (n = 1), cardiopulmonary bypass strategy (n = 10), packed red blood cells transfusion (n = 1), intra-aortic balloon pump (n = 1), and pulmonary artery catheterization (n = 1)], and postoperative care [red blood cell transfusion (n = 1)] (Table 2). Of the 12 identified factors, we synthesized 11 mechanisms from 22 articles for their effect on mortality (Table 3).

Intermediary outcomes of in-hospital mortality

In total, nine articles reported intermediary outcome of in-hospital mortality after CABG. Factors in these articles were grouped to treatment delay [surgical delay (n = 1); timing of surgery n = 4)] and complications [presence of factor (n = 1), hyperthermia (n = 1), hypothermia (n = 1), early postoperative stroke (n = 1), pulmonary artery temperature on ICU admission (n = 1)] (Table 2). Of the seven identified factors, we synthesized 1 mechanism from 1 article for its effect on mortality (Table 3).

Case-mix factors of in-hospital mortality

In total, 36 articles reported on case-mix factors of in-hospital mortality after CABG. Factors in these articles were grouped to sociodemographic factors [sex (n = 4), Native American status (n = 1), Medicaid insurance or uninsured status (n = 1)], health risks [body mass index (n = 3)],

disease characteristics [CAD diffuseness (n = 1)], disease history [prior PCI (n = 3)], comorbidity burden [atrial fibrillation (n = 1), diabetes (n = 1), dialysis-dependent renal failure (n = 2), metabolic syndrome (n = 1), non-dialysis-dependent renal failure (n = 1), peripheral vascular disease (n = 1), peritoneal dialysis (n = 2), preoperative neurological events (n = 1), preoperative reduced ejection fraction (n = 1), QT prolongation (n = 1), renal dysfunction (n = 1), renal insufficiency (n = 1), right ventricular systolic dysfunction (n = 1)], and operative risks [Cockcroft-Gault formula to evaluate renal function (n = 1), C-reactive protein (n = 1), forced expiratory volume 1 (n = 1), left atrial expansion index (n = 1), red cell distribution width (n = 1), REMEMBER score (n = 1), serum creatinine (n = 1), white blood cell count (n = 1)] (Table 2). Of the 27 identified factors, we synthesized 18 mechanisms from 24 articles for their effect on mortality (Table 3).

Discussion

Summary of evidence

The purpose of this scoping review was to map factors of in-hospital mortality after CABG to the augmented Donabedian framework of quality improvement, and to synthesize mechanisms for their effect on mortality. We selected factors of mortality reported in 83 articles and sorted them into 14 groups according to common attributes. We mapped the groups to case-mix, structure, process, and outcome elements of the augmented Donabedian framework for quality of care (Fig. 2). The majority (44%) of articles reported on the characteristics of patients, their disease, and their health status. Factors related to care processes were reported in 33% of the articles, and structures 13% of the articles. We synthesized 33 mechanisms for factor association on mortality.

Fig. 2
figure 2

Post-operative mortality factor groups within the augmented Donabedian framework

Full size image

These findings suggest that the patient’s demographic characteristics, their social determinants, health risks, disease characteristics, disease history, comorbidity burden, and operative risks are more frequently assessed risk factors of in-hospital mortality. However, factors in these groups are largely unsuitable for quality improvement given the time available for intervention between surgery and in-hospital death, and therefore may be considered for risk stratification of patients, as suggested by Shroyer [17].

Our results showed process factors of mortality were identified in less than half of the reviewed articles, and structural factors in approximately one in ten reviewed articles. This may be a function of data collection practices, if care process documentation is not translated into records in the cardiac surgery database. Equally, data on structural factors of mortality may not be collected at all if the database is for a single institution or if the registry focus is more epidemiological than one that supports health services research. Therefore, an opportunity may exist for cardiac surgery database managers to incorporate collection of information on both structural and process of care factors into their databases.

An interesting finding of our scoping review was the number of studies reporting on the use of cardiopulmonary bypass strategy—specifically on-pump CABG compared to off-pump CABG—as a factor of mortality, with several papers suggesting mechanisms for the effect [28, 52, 55, 62, 71, 78, 84, 85]. Multiple randomized controlled trials [105,106,107,108] have shown no difference in mortality at 30 days between the two approaches, with a five-year extension to the CORONARY trial showing no long-term difference [109]. This may be due to differences in the internal validity of the methodological approaches. For example, observational studies cannot control for unobserved confounding. Alternatively, it may be due to differences in the external validity of the approach whereby observational studies better reflect the entire population versus those that are suitable for enrollment into randomized controlled trials.

Shroyer [110] wrote that outcomes indirectly provide information on potential challenges, and do not identify specific actions to be taken. In response, we extracted and synthesized mechanisms for the effect of 52 factors of mortality from 83 articles, approximately 63% of those reviewed. While these results provide insight into the effect of the factors, it offers limited targets for improvement given only 15 mechanisms were identified for factors mapped to structures and processes of care groups, and modifiable case-mix factors, such as BMI, may not be so during the period between surgery and in-hospital mortality. Thus, initiatives to improve care quality will have limited number of factors and information from which to guide their intervention design.

Limitations

We did not select studies published prior to 2000 to minimize the potential biasing effect of surgical advancements and changes in delivery of coronary artery bypass grafting [1]. This may have led to an underestimation of the extent of prognostic factors of mortality. We limited our search to works published in English and in PubMed, CINAHL, or EMBASE. Additional studies may be non-English and/or published in databases not included in our search strategy. This may have led to an overestimation of the extent of prognostic factors of mortality as positive results are more likely to be published and reported in English language studies [111]. We excluded randomized controlled trials as their findings do not necessarily reflect mortality after coronary artery bypass grafting following usual care. While this may have led to exclusion of potentially relevant literature, observational data reflects real-world mortality and can better inform quality improvement efforts. We excluded studies that used a composite measure of complications and mortality. We excluded studies which did not complete regression analysis as we used regression effect estimates to enable identification of the direction of the reported association [112]. Further, we limited our search strategy to studies of isolated coronary artery bypass grafting due to different projected outcomes across procedures for coronary revascularization [113]. The results are therefore not generalizable to other revascularization procedures. We also limited our search to mortality in hospital to reduce the likelihood of unobserved factors confounding mortality outcomes after discharge from hospital. With reductions in acute length of stay, it is possible we underestimated the extent of prognostic factors of in-hospital mortality [114]. We used statistical significance to identify the presence of an association between the factor and mortality; this work does not describe the strength of the association which may further inform which factors to target for intervention. When selecting factors, we reported the presence of an association, not the strength of the association. Finally, we did not assess the quality of the reviewed articles per the scoping review framework [115]

Conclusion

Previous research reports numerous factors of post-operative mortality in patients undergoing CABG. This evidence has not been mapped to the conceptual framework of quality improvement.

We identified 52 factors of mortality reported in 83 articles and mapped them to 14 groups of contributing to mortality onto the augmented Donabedian framework for quality of care, which includes case mix, structure, process, and intermediary outcomes. Most factors included proposed mechanisms for their mortality effects. The majority of factors reported were immutable factors, related to characteristics of patients, their disease and their pre-operative health status. Modifiable factors related to care structures and intermediary outcomes were least reported, with factors related to care processes reported in only one-third of the articles. Therefore, there are limited evidence-based opportunities to improve mortality that will reduce variation in mortality after coronary artery bypass graft surgery. Future studies should consider studying modifiable factors that may be intervened upon to improve mortality directly or through their modifiable mechanism.