BACKGROUND:Bioreactance estimates cardiac output in a non-invasive way. We evaluated the ability of a bioreactance device (NICOM®) to estimate cardiac index (CI) and to track relative changes induced by volume expansion. METHODS:In 48 critically ill patients, we measured CI estimated by the NICOM® device (CINicom) and by transpulmonary thermodilution (CItd, PiCCO2™ device) before and after a 500 ml saline infusion. Before volume expansion, we performed a passive leg raising (PLR) test and measured the changes it induced in CINicom and in pulse contour analysis-derived CI. RESULTS:Considering the values recorded before PLR and before and after volume expansion (n=144), the bias (lower and upper limits of agreement) between CItd and CINicom was 0.9 (-2.2 to 4.1) litre min(-1) m(-2). The percentage error was 82%. There was no significant correlation between the changes in CItd and CINicom induced by volume expansion (P=0.24). An increase in CI estimated by pulse contour analysis >9% during the PLR test predicted fluid responsiveness with a sensitivity of 84% (95% confidence interval 60-97%) and a specificity of 97% (95% confidence interval 82-100%). The area under the receiver operating characteristic curve constructed to test the ability of the PLR-induced changes in CINicom in predicting fluid responsiveness did not differ significantly from 0.5 (P=0.77). CONCLUSIONS:The NICOM® device cannot accurately estimate the cardiac output in critically ill patients. Moreover, it could not predict fluid responsiveness through the PLR test.