Level of consciousness at the end of life in critically ill patients is poorly characterized. We report a case series of seven patients who were neurologically intact before the decision to withdraw care due to extensive systemic critical illness. As part of our end-of-life care protocol, bispectral index (BIS) monitor (Aspect Medical Systems, Newton, MA) or SEDline (Hospira, Lake Forest, IL) monitoring devices are placed on each patient to ensure adequate comfort. Both monitoring systems use an integer-based system (BIS or PSI, respectively) to reflect the level of consciousness/effect of anesthesia. In each case, loss of blood pressure, as monitored by indwelling arterial line, was followed by a decline is BIS/PSI activity followed by a transient spike in BIS/PSI activity that approached levels normally associated with consciousness. This spike in electroencephalogram (EEG) activity had short duration and the activity then declined to a level of activity associated with burst suppression. In one case of a patient who had a SEDLine device, we were able to capture and analyze the raw EEG signal, and confirm that the EEG waveform was not artifact, and in fact a high frequency waveform was present during the spike activity. We speculate that this level of BIS/SEDline activity is related to the cellular loss of membrane polarization due to hypoxemia. We further speculate that since this increase in electrical activity occurred when there was no discernable blood pressure, patients who suffer "near death" experiences may be recalling the aggregate memory of the synaptic activity associated with this terminal but potentially reversible hypoxemia.

PurposeSecond-generation FloTrac software has been shown to reliably measure cardiac output (CO) in cardiac surgical patients. However, concerns have been raised regarding its accuracy in vasoplegic states. The aim of the present multicenter study was to investigate the accuracy of the third-generation software in patients with sepsis, particularly when total systemic vascular resistance (TSVR) is low.MethodsFifty-eight septic patients were included in this prospective observational study in four university-affiliated ICUs. Reference CO was measured by bolus pulmonary thermodilution (iCO) using 3–5 cold saline boluses. Simultaneously, CO was computed from the arterial pressure curve recorded on a computer using the second-generation (COG2) and third-generation (COG3) FloTrac software. CO was also measured by semi-continuous pulmonary thermodilution (CCO).ResultsA total of 401 simultaneous measurements of iCO, COG2, COG3, and CCO were recorded. The mean (95%CI) biases between COG2 and iCO, COG3 and iCO, and CCO and iCO were −10 (−15 to −5)% [−0.8 (−1.1 to −0.4) L/min], 0 (−4 to 4)% [0 (−0.3 to 0.3) L/min], and 9 (6–13)% [0.7 (0.5–1.0) L/min], respectively. The percentage errors were 29 (20–37)% for COG2, 30 (24–37)% for COG3, and 28 (22–34)% for CCO. The difference between iCO and COG2 was significantly correlated with TSVR (r2 = 0.37, p < 0.0001). A very weak (r2 = 0.05) relationship was also observed for the difference between iCO and COG3.ConclusionsIn patients with sepsis, the third-generation FloTrac software is more accurate, as precise, and less influenced by TSVR than the second-generation software.Electronic supplementary materialThe online version of this article (doi:10.1007/s00134-010-2098-8) contains supplementary material, which is available to authorized users.

Decreased levels of serum albumin concentration, increased A-a gradient, and presence of active cancer predict which patients who are admitted to the ICU will develop ARF.