Extravascular lung water index: Diagnostic accuracy and relation to lung injury and mortality in patients with shock

INTRODUCTION. The diagnosis of acute lung injury may be more robust if more accurate physiological markers can be identified. Extravascular lung water index (EVLWI) may be useful and has been shown to correlate with respiratory function and mortality in patients with sepsis and ARDS. Whether this applies to a wider population, and which index performs best, are unclear. OBJECTIVES. We hypothesized that EVLWI correlates with respiratory function and mortality in patients with documented systemic inflammation and shock. We investigated EVLW indexed to actual and predicted body weight, and pulmonary blood volume. We investigated the diagnostic accuracy of EVLWI for lung injury. METHODS. In 51 patients with shock and SIRS, EVLWI was measured within 6 h of ICU admission and indexed to actual weight (EVLWI/ABW), predicted body weight (EVLWI/ PBW) and pulmonary blood volume (EVLWI/PBV). Relationships to lung injury and ICUmortality were investigated. Positive and negative likelihood ratios, pre- and post-test odds and ROC curves were calculated. RESULTS. EVLWI was higher among patients with lung injury and was significantly correlated with respiratory parameters. EVLWI/ABW was higher among non-survivors and gave the best positive likelihood ratios for diagnosing ALI/ARDS. In contrast, EVLWI/PBV gave better diagnostic value for severe lung injury according to Murray's LIS criteria. The post-test odds for ALI and ARDS increased threefold when using EVLWI/ABW as a bedside test. The post-test odds of severe lung injury increased eightfold using EVLWI/PBV. EVLWI/ABW and EVLWI/PBV generated the best ROC curves for mortality prediction with a sensitivity of 68% and specificity of 63-72%. CONCLUSIONS. EVLWI was associated with degree of lung injury, regardless of the index used, supporting its usefulness as a bedside indicator for disease severity. EVLWI/PBV and EVLW/ABW gave the best diagnostic accuracies for the diagnosis of lung injury, and generated the bestROCcurves for mortality prediction.EVLWI/ABWwas significantly increased in non-survivors. Further studies are needed to confirm the additional value of EVLWI for the early identification of lung injury