The liver, cardiac surgery and EuroSCORE

The article by Ostovar et al. [1] addresses the fact that liver cirrhosis (and, indeed, liver dysfunction in general) do not feature in any of the EuroSCORE risk models [2, 3]. In this observational study, they looked at 218 cardiac surgical patients with 3 levels of liver cirrhosis: Child-Pugh A, B and C. They compared early mortality for these patients with that of a contemporaneous cohort of 5764 cardiac surgical patients with no cirrhosis. Their results are powerful and compelling: early mortality for the Child A cirrhosis group was 8.9% (virtually identical to that of non-cirrhotic patients) but, in Child B and Child C groups, the mortality was, respectively, 52% and 100%. I hope that the authors will forgive me for gently questioning the mortality figure for the non-cirrhotic patients: it appears to be significantly higher than EuroSCORE II predicts and I hope and presume that these results will have been reviewed at the institution. It is also a little surprising that the prevalence of cirrhosis in the cardiac surgical patients in this study appears to be 3.8% (the equivalent figure for all American adults is 0.27%) [4]. It is, of course, difficult to compare this prevalence with available population statistics where liver diseases are often treated as a group and there is no comparator population that matches the age, sex and health profile of cardiac surgical patients. None of this, however, in any way diminishes the importance of these study findings. All cardiac surgeons and cardiologists know that liver disease is an important risk factor in cardiac surgery, for all the reasons quoted in the article and probably for other reasons that are not yet fully understood. Nevertheless, the impact of this risk factor on cardiac surgery survival appears to be much more severe than most surgeons, myself included, had ever suspected. As such, these findings alone should be seen as a forewarning to any surgeon contemplating major cardiac surgery on a patient with cirrhosis classes B and C. Cirrhosis and other forms of liver disease, in general, are conspicuously absent from the EuroSCORE risk model. When EuroSCORE II data were being collected, we included serum albumin as a marker for liver disease, but our multiple attempts to incorporate that factor into the model did not improve model performance or stability and were unsuccessful. This article used the Child-Pugh classification instead as the liver disease risk factor and substantially improved EuroSCORE’s performance, so it is something that should be taken into account when data are being collected for future iterations of the risk model. As of now, the 2 strongest contenders for future inclusion in EuroSCORE are frailty and liver disease. Of course, the requirements for inclusion of a risk factor in a model are its universal availability (preferably by objective measurement), the weight of its impact and the effect that its inclusion has on the robustness and stability of the model. Both frailty and liver disease did not make the mark in EuroSCORE II for one or more of these reasons. By proposing the Child-Pugh classification, Ostovar et al. may have solved that problem for liver disease. Later this year, data collection will begin for EuroSCORE 3, and we shall take this article seriously when we come to the selection of risk factors. In the meantime, surgeons contemplating major cardiac surgery on a patient with Child B or C cirrhosis should only proceed with the utmost caution and after the most careful consideration of the risk of surgery, that of no surgery and that of alternative treatment modalities.