ECMO Predictors of Mortality: A 10-Year Referral Centre Experience

Abstract

Background: Extracorporeal membrane oxygenation (ECMO) is a specialised life support modality for patients with refractory cardiac or respiratory failure. Multiple studies strived to evaluate the benefits of ECMO support, but its efficacy remains controversial with still inconsistent and sparse information. Methods: This retrospective analysis included patients with ECMO support, admitted between January 2010 and December 2019 at a tertiary university ECMO referral centre in Austria. The primary endpoint of the study was overall all-cause three-month mortality with risk factors and predictors of mortality. Secondary endpoints covered the analysis of demographic and clinical characteristics of patients needing ECMO, including incidence and type of adverse events during support. Results: In total, 358 patients fulfilled inclusion criteria and received ECMO support due to cardiogenic shock (258, 72%), respiratory failure (88, 25%) or hypothermia (12, 3%). In total, 41% (145) of patients died within the first three months, with the median time to death of 9 (1–87) days. The multivariate analysis identified hypothermia (HR 3.8, p < 0.001), the Simplified Acute Physiology Score III (HR 1.0, p < 0.001), ECMO initiation on weekends (HR 1.6, p = 0.016) and haemorrhage during ECMO support (HR 1.7, p = 0.001) as factors with higher risk for mortality. Finally, the most frequent adverse event was haemorrhage (160, 45%) followed by thrombosis. Conclusions: ECMO is an invasive advanced support system with a high risk of complications. Nevertheless, well-selected patients can be successfully rescued from life-threatening conditions by prolonging the therapeutic window to either solve the underlying problem or install a long-term assist device. Hypothermia, disease severity, initiation on weekends and haemorrhage during ECMO support increase the risk for mortality. In the case of decision making in a setting of limited (ICU) resources, the reported risk factors for mortality may be contemplable, especially when judging a possible ECMO support termination.