OUP user menu

The logistic EuroSCORE

F Roques, P Michel, A.R Goldstone, S.A.M Nashef
DOI: http://dx.doi.org/10.1016/S0195-668X(02)00799-6 882 First published online: 1 May 2003

Dear Sir,

The European System for Cardiac Operative Risk Evaluation (EuroSCORE) identifies a number of risk factors which help to predict mortality from cardiac surgery.1 The predicted mortality (in percent) is calculated by adding the weights assigned to each factor. Since its initial publication in 1999, EuroSCORE has been widely used in Europe and elsewhere and has been the subject of several studies. Most of these studies compared the predictive ability of the EuroSCORE to previously described systems2,3 or to locally derivedmodels.4 One original work used it as a tool in order to assess the intra-institutional benefit in switching from conventional surgery under cardiopulmonary bypass to the off-pump approach in coronary surgery.5 In general, EuroSCORE was found to be an easy tool for inter-institutional comparison with good or excellent predictive ability. Nevertheless, many observers noted a trend to an underestimation of the operative risk in very high-risk patients, and it has been suggested that full statistical comparison to other systems might be difficult since comprehensive information on the logistic regression equation of the score was neverpublished.

EuroSCORE was initially designed to be a user-friendly system, in the hope of encouraging as many units as possible to embark on programmes of risk-adjusted quality monitoring. In this setting, although derived from a logistic regression methodology, only the simple additive version of the score was originally published. This score could be easily calculated at the bedside4,5 and could therefore be used widely in Europe even in hospitals with little information technology. Today, European cardiologists and cardiac surgeons are demonstrating a growing interest in quality control and have access to ever better information technology resources. Some may therefore wish to use a more sophisticated risk model than the simple additive EuroSCORE. For this reason, we seek to make public the full details of the EuroSCORE logistic regression equation (Table 1). Using the same risk factors, the logistic regression version of the score (the ‘logistic EuroSCORE’) can be calculated. For a given patient, the logistic EuroSCORE which is the predicted mortality according to the logistic regression equation, can be achieved with the following formula:Mathwhere β0is the constant of the logistic regression equation (see Table 1) and βi is the coefficient of the variable Xi in the logistic regression equation provided in Table 1. Xi=1 if a categorical risk factor is present and 0 if it is absent. For age, Xi=1 if patient age <60; Xi increases by one point per year thereafter (ie: age 59 or less Xi=1; age 60 Xi=2; age 61 Xi=3 and so on).

View this table:
Table 1

Logistic regression model of EuroSCORE in the 1995 pilot study

Variablesβ Coefficient
Age (continuous)0.0666354
Serum creatinine >200μmol/l0.6521653
Extracardiac arteriopathy0.6558917
Pulmonary disease0.4931341
Neurological dysfunction0.841626
Previous cardiac surgery1.002625
Recent myocardial infarct0.5460218
LVEF 30–50%0.4191643
LVEF <30%1.094443
Systolic pulmonary pressure >60mmHg0.7676924
Active endocarditis1.101265
Unstable angina0.5677075
Emergency operation0.7127953
Critical preoperative state0.9058132
Ventricular septal rupture1.462009
Other than isolated coronary surgery0.5420364
Thoracic aortic surgery1.159787
Constant β0−4.789594
  • LVEF, left ventricular ejection fraction; full definition of these variables are published1 and can be seen on-line (http://www.euroscore.org).

Cardiologists and cardiac surgeons will be relieved to note that they do not have to face this complex calculation with every patient; a risk calculator (additive and logistic) can be used or downloaded easily from the EuroSCORE website (http://www.euroscore.org).

The additive EuroSCORE is simple, well validated, user-friendly and works at the bedside without specialised equipment. Because of its additive properties, it will tend to underestimate risk in some very high risk groups. The logistic EuroSCORE is more suitable for individual risk prediction in very high risk patients and will facilitate further sophisticated study into the field of risk. Within Europe and elsewhere in the world there are hospitals with rudimentary, if any, data collection and others, where data are abundant and the study of risk is a specialised area of advanced investigation. It can therefore be argued that there is currently a place for both simple as well as sophisticated risk tools in this rapidly evolving field.