Estimated versus Actual Oxygen Consumption in Patients Supported with LVADs

Purpose Accurately measured cardiac output (CO) is critical to medical decision making in heart failure patients. The most recent OPTN listing criteria incorporate CO in determining allocation priority. CO is frequently calculated using the Fick equation, in which the key variable of oxygen consumption (VO2) can be directly measured, but is more commonly estimated. The accuracy of these VO2 estimations have been questioned in heart failure patients and have not been validated in LVAD patients. We sought to determine the comparative accuracy of estimated versus measured VO2 in a cohort of LVAD patients using three common formulae. Methods Data were prospectively collected in consecutive LVAD patients who underwent right heart catheterization with direct measurement of resting VO2 using an MGC Diagnostics Ultima CardiO2® gas-exchange analysis cart. We estimated VO2and Fick CO using the formulae proposed by Dehmer et al., LaFarge and Miettinen, and Bergstra et al., and compared these values with directly measured VO2, Fick CO, and thermodilution CO. Results Four male patients with HeartMate 3® LVADs underwent right heart catheterization with VO2 measurement. The mean differences in estimated versus measured VO2 and CO were smallest with the formulae proposed by Dehmer et al. and LaFarge and Miettinen, which both modestly underestimated directly measured CO (-0.2 ± 0.3). The error in calculated CO was substantial when utilizing the Bergstra et al. formula (0.5 ± 0.4) and the thermodilution method (-0.8 ± 0.2). Conclusion Resting oxygen consumption in LVAD patients, as directly measured by gas-exchange analysis, can be estimated within an acceptable degree of error by the formulae proposed by Dehmer et al. and LaFarge and Miettinen. By comparison, the assumed VO2 by the Bergstra et al. formula appears less accurate in LVAD patients and the degree of error in calculated Fick CO using this equation may be unacceptable for clinical decision making.