Oxygen extraction in patients with sepsis and heart failure: another look at clinical studies.

We collected all complete sets of measurements of cardiac index and oxygen-derived variables available in the recent literature (1975-1991; computerised Medline search) on critically ill patients with sepsis (n=21 studies), septic shock (n=20 studies) or severe heart failure (n=13 studies). For each study, the mean value for cardiac index, oxygen delivery (DO 2), oxygen uptake (VO 2), oxygen extraction ratio (O 2ER) and lactate concentration (when available) were analysed together with mortality rates. There was a significant relationship between VO 2 and DO 2 for the studies on patients with severe heart failure (r=0.79, p less than 0.001) or septic shock (r=0.55, p less than 0.01), but not in patients with sepsis (r=0.3, p=NS). As expected, O 2ER was higher in the studies in heart failure (31 to 50%) than in septic shock (25 to 45%) or sepsis (19 to 40%). When compared to the studies on patients with sepsis, the studies in septic shock included patients with lower DO 2 (481+/-89 vs 539+/-79 ml/min.M 2, p=0.032) and higher O 2ER (32.5+/-5.2 vs 27.8+/-5.0%, p less than 0.01) but similar VO 2 (143+/-29 vs 143+/-19 ml/min.M 2, p=NS). In a cardiac index/O 2ER diagram where a line of reference represents proportional changes in cardiac index and O 2ER from normal values at rest, all studies in heart failure fell below the line of reference. The majority of studies in sepsis (18/21) fell above the line of reference. Interestingly, only nine of the 20 studies in septic shock fell above this line. The review of the 14 studies in septic shock, in which O 2ER and blood lactate were reported, showed a direct relationship between these two variables (r=0.68, p less than 0.01), but no relationship between cardiac index and blood lactate (r=0.37, p=NS). This analysis suggests that patients with septic shock tend to have a lower cardiac index and higher O 2ER than septic patients who are haemodynamically stable. Furthermore, the positive relationship between O 2ER and lactate suggests that, despite the alterations in oxygen extraction capabilities in severe sepsis, mean O 2ER may be higher in the most severe cases of septic shock. This could reflect an attempt to maintain VO 2 when DO 2 is insufficient. The prevailing opinion that 0 2ER is a meaningless variable in septic shock needs to be reassessed.