The ejection fraction and the cardiac index in the perioperative period of coronary artery bypass grafting: what is worth relying on?

Rationale: Ejection fraction (EF) has been recognized as a key echocardiographic parameter describing the systolic function of the heart. Nevertheless, its accuracy for the assessment of the hemodynamic status in the perioperative period of cardiac surgery remains a debate. Aim: To compare the EF values assessed by different techniques and to map those against the left ventricular systolic function parameters in the perioperative period of the off-pump coronary artery bypass grafting (OPCABG). Methods: We performed a post hoc analysis of two consequential randomized studies. EF was assessed with echocardiography in 115 patients in the perioperative period of OPCABG and its changes over time at the following timepoints: before admission to the hospital (EFscreening), on the day before surgery (EFsimpson) (transthoracic approach), before sternotomy, at the end of the surgery (transesophageal echocardiography, TEE), as well as at the end of the first postoperative day (POD1) and before discharge from the hospital (transthoracic approach). Preoperatively and at the end of POD1, the EF values were compared with those of the global longitudinal strain (GLS) and with the EF measured automatically (EFauto, QLAB 10.0), as well as with hemodynamic parameters measured by thermodilution (TD), such as cardiac index (CI) and stroke volume index (SVI). Results: There were no postoperative changes in the EFscreening and EFsimpson values. EFauto decreased from 48.4 ± 6.4 to 41.6 ± 7.3% (p 0.001), while EFtee increased from 49.7 46.0; 57.0 to 53.0 46.1; 58.1% (p = 0.047) and was associated with a decrease in GLS from -14.6 ± 2.5 to -11.7 ± 2.6% (p 0.001). The bias in the measurement of EFsimpson and EFauto assessed a day before surgery and on POD1 according to the Bland-Altman test was 29.3% and 34.0%. There was a correlation between GLS and EFauto preoperatively (rho = -0.791, p 0.001) and on the POD1 (rho = -0.723, p 0.001), while EFscreening and EFsimpson did not show such a correlation. There was no correlation of EF or GLS with CI, as well. The bias in the measurement of the SVItd and SVIecho values was 11% (rho = 0.301, p = 0.001). Conclusion: The echocardiographic assessment of the EF in the perioperative period of OPCABG does not characterize the systolic function of the heart reliably and depends on the method of its measurement. These limitations of echocardiographic parameters require their critical evaluation, including a comparison with thermodilution variables.