Systematic review including re-analyses of 1148 individual data sets of central venous pressure as a predictor of fluid responsiveness

Central venous pressure (CVP) has been shown to have poor predictive value for fluid responsiveness in critically ill patients. We aimed to re-evaluate this in a larger sample subgrouped by baseline CVP values. In April 2015, we systematically searched and included all clinical studies evaluating the value of CVP in predicting fluid responsiveness. We contacted investigators for patient data sets. We subgrouped data as lower (< 8 mmHg), intermediate (8-12 mmHg) and higher (> 12 mmHg) baseline CVP. We included 51 studies; in the majority, mean/median CVP values were in the intermediate range (8-12 mmHg) in both fluid responders and non-responders. In an analysis of patient data sets (n = 1148) from 22 studies, the area under the receiver operating curve was above 0.50 in the < 8 mmHg CVP group [0.57 (95 % CI 0.52-0.62)] in contrast to the 8-12 mmHg and > 12 mmHg CVP groups in which the lower 95 % CI crossed 0.50. We identified some positive and negative predictive value for fluid responsiveness for specific low and high values of CVP, respectively, but none of the predictive values were above 66 % for any CVPs from 0 to 20 mmHg. There were less data on higher CVPs, in particular > 15 mmHg, making the estimates on predictive values less precise for higher CVP. Most studies evaluating fluid responsiveness reported mean/median CVP values in the intermediate range of 8-12 mmHg both in responders and non-responders. In a re-analysis of 1148 patient data sets, specific lower and higher CVP values had some positive and negative predictive value for fluid responsiveness, respectively, but predictive values were low for all specific CVP values assessed.