Original Article
Comparison of Four Risk Scores for Contemporary Isolated Coronary Artery Bypass Grafting

https://doi.org/10.1016/j.hlc.2013.12.001Get rights and content

Abstract

Background

EuroSCORE and the Society of Thoracic Surgeons’ (STS) Score have been the most widely used risk scores for cardiac surgery. The revised EuroSCORE II and the AusSCORE, based on an Australasian population, were recently developed. We compared the prognostic utility of these four scores for mortality as well as morbidity in patients undergoing isolated coronary artery bypass grafting (CABG).

Methods

The scores were retrospectively calculated for isolated CABG patients at Auckland City Hospital during July 2010-June 2012. Discrimination and calibration of outcomes were assessed.

Results

818 patients were followed for 1.6+/-0.6 years. Mortality at 30 days was 1.6% and 2.9% on follow up. Median predicted 30 day mortality (Interquartile range) for EuroSCORE I were 2.8% (1.6%, 5.2%), EuroSCORE II 1.6% (1.0%, 2.8%), STS Score 2.3% (1.3%, 4.5%) and AusSCORE 0.5% (0.2%, 1.1%). C-statistics and Hosmer-Lemeshow test p-values for these scores for 30-day mortality were Euro score I 0.675 (95%CI 0.531-0.819)/0.061, EuroSCORE II 0.642 (0.503-0.780)/0.150, STS Score 0.641 (0.507-0.775)/0.243 and AusSCORE 0.661 (0.516-0.807)/0.420.

Only EuroSCORE I and STS scores were significant for predicting mortality at follow-up (c = 0.639 and 0.666). All scores predicted composite morbidity. C-statistics were EuroSCORE I 0.678, EuroSCORE II 0.634, STS score 0.584 and AusSCORE 0.645.

Conclusion

EuroSCORE II, STS Score and AusSCORE had slightly improved calibration but similar discrimination for 30-day mortality compared to EuroSCORE I.

Revision of risk models to fit contemporary surgical outcomes is important, but there may only be modest room for improvement in discrimination.

Introduction

Several operative risk scores for cardiac surgery have been developed in the last few decades including the Parsonnet Score, [1] EuroSCOREs [2], [3] and Society of Thoracic Surgeon's (STS) score. [4] EuroSCORE I was developed from a European cohort of 14,781 patients having cardiac surgery during 1995 for 30-day mortality, and published as an additive model in 1999 [2] and logistic model in 2003. [3] The STS score was developed to predict operative morbidity and was derived from an American cohort of 774,881 isolated coronary artery bypass grafting (CABG) patients during 2002-2006 and published in 2008. [4]

Despite the early validation of EuroSCORE I in large international populations, [5], [6] more recent studies found the score over-estimated operative mortality, probably because of improving operative and peri-operative management. [7], [8] In Australasian populations characterised by significant ethnic diversity, the EuroSCORE I also over-estimated operative mortality. [9] The AusSCORE was published in 2009 from 11,823 patients undergoing isolated CABG during 2001-2005 from the Australasian Society of Cardiac and Thoracic Surgeon's (ASCTS) database to predict 30-day mortality. [10] To date, there are no studies assessing its external validity.

More recently as a project to revise the original EuroSCORE to fit contemporary cohorts, the EuroSCORE II was developed from an international cohort of 22,381 patients undergoing cardiac surgery during 2010 and published in 2012. [11] Studies which have assessed the external validity of this new score have reported mixed results with EuroSCORE II performing better [12], [13], [14], [15] or similar [16], [17] to EuroSCORE I.

EuroSCORE II, STS Score and AusSCORE have not been directly compared for CABG, or assessed in Australasian cohorts. In addition the comparative value of the different scores for predicting mortality beyond 30 days is uncertain. Our objective was to compare the predictive efficacy of logistic EuroSCORE I, EuroSCORE II, STS Score and AusSCORE for morbidity and mortality at 30 days and longer follow-up after isolated CABG.

Section snippets

Patient selection and data collection

Ethics approval of this study was obtained from our institution's ethics review committee. Consecutive patients undergoing isolated CABG without concomitant valve surgery at Auckland City Hospital were included from July 2010 to June 2012. Relevant clinical characteristics were collected from computerised records. Logistic EuroSCORE I, [3] EuroSCORE II, [11] STS Score [4] and AusSCORE, [10] were retrospectively calculated from all patients using available data.

The EuroSCORE II definitions were

Results

Table 1 presents the baseline characteristics of the study population. Mean age was 64.5+/-10.0 years and 20.2% (168) were female. Mean follow-up was 1.6+/-0.6 years. The median predicted 30-day mortality (interquartile range IQR) for EuroSCORE I was 2.8% (1.6%, 5.2%), EuroSCORE II 1.6% (1.0%, 2.8%), STS Score 2.3% (1.3%, 4.5%) and AusSCORE 0.5% (0.2%, 1.1%). Table 2 lists the operative variables and post-operative outcomes.

Discussion

Our study shows that EuroSCORE I, EuroSCORE II, STS Score and AusSCORE were all able to discriminate outcomes after CABG with modest accuracy and varying strengths. EuroSCORE II, STS Score and AusSCORE had slightly better calibration than EuroSCORE I for 30-day mortality, but discrimination for outcomes were not superior to EuroSCORE I.

EuroSCORE I had been a success for over a decade since its introduction as the primary international risk score for operative mortality prediction, but recent

Conclusion

In a contemporary cohort of patients undergoing isolated CABG the EuroSCORE II, STS Score and AusSCORE had modest improvements in calibration for 30-day mortality, as compared with EuroSCORE I. All four scores predicted some but not all post-operative complications. Revision of risk models to fit contemporary surgical outcomes is important for calibration, but the room for improvement for discriminating adverse outcomes may be limited. Given the modest C-statistics found in our analysis, there

Disclosures

None of the authors have any disclosures.

Acknowledgements

No funding was utilised in this research. We would like to thank all our colleagues affiliated with the Green Lane Cardiovascular Service for their contributions and support and Charlene Nell, Desktop Support Administrator, for excellent secretarial assistance.

References (24)

  • S.A. Nashef et al.

    Validation of European System for Cardiac Operative Risk Evaluation (EuroSCORE) in North American cardiac surgery

    Eur J Cardiothorac Surg.

    (2002)
  • F. Bhatti et al.

    The logistic EuroSCORE in cardiac surgery: how well does it predict operative risk?

    Heart.

    (2006)
  • Cited by (22)

    • Evidence-Based Determination of Cut-Off Points for Increased Cardiac-Surgery Mortality Risk With EuroSCORE II and STS: The Best-Performing Risk Scoring Models in a Single-Centre Australian Population

      2022, Heart Lung and Circulation
      Citation Excerpt :

      The STS morbidity model over predicted DSWI and NRF. Our observations that AES and LES overestimated mortality, whereas ASMR and STS underestimated mortality compared to observed mortality, are consistent with results from some previous studies [2,12,13,15,17,21,33,34] but not others [1,14,16]. Discrepancy in RSM calibration can be attributed to differences in population characteristics, procedural subsets, and risk strata and can vary with time and location of application [1,2,8,10,13–16,18,22,23].

    • Machine learning to predict mortality risk in coronary artery bypass surgery

      2021, Cardiovascular and Coronary Artery Imaging: Volume 1
    • State-of-the-Art Coronary Artery Bypass Grafting: Patient Selection, Graft Selection, and Optimizing Outcomes

      2019, Interventional Cardiology Clinics
      Citation Excerpt :

      The latter scoring system also provides a calculated risk of stroke, renal failure, sternal wound infection, and length of stay.13 The predictive ability of these 2 models performs equally in the setting of isolated CABG.14 However, other comorbidities, such as pulmonary hypertension, liver disease, previous chest radiation, and the frailty status of the patient, should be included as significantly increasing the surgical risk.15

    View all citing articles on Scopus
    View full text