LAUSR

To the Editor: We read with great interest the review of Long et al. (1) entitled ‘‘Does respiratory variation in inferior vena cava diameter predict fluid responsiveness: a systematic review and meta-analysis,’’ published in this journal, which provides a critical overview on the controversial ability of the respiratory changes of inferior vena cava (IVC) to predict fluid responsiveness in the critically ill. However, the conclusion of their limited clinical utility in spontaneously breathing patients may be revised, considering the following points. First, two of the five studies that were selected for the review focused on spontaneously breathing patients used inaccurate techniques to assess fluid responsiveness in the critically ill (i.e., noninvasive blood pressure (2) and bioreactance (3)) and should probably not be taken into account. Second, it should be noted that Lanspa et al. (4) discussed two threshold values of caval index in their study, >15% and >50%. The sensitivity and specificity reported in Table 3 of the review and used for the calculation correspond to the threshold value of>15%, when the use of the threshold value of >50% would have been more homogenous compared with those of Airapetian et al. (>42%) and Muller et al. (>40%). Third, our team recently found that IVC respiratory variations assessed in 90 spontaneously breathing patients with septic circulatory failure predicted fluid responsiveness with a negative predictive value (NPV) of 74%, and a positive predictive value (PPV) of 88% at a threshold value of >31% (5). When these results are pooled with those of the three other reliable studies (4, 6, 7), the caval index predicts fluid responsiveness with NPVof 69%, and a PPV of 86% (Table 1). Thus, a high collapsibility of the IVC is often associated with fluid responsiveness, but as discussed in the review, low values cannot rule out a positive response to fluid infusion. Yet, the main concern for clinicians is to avoid useless or deleterious volume expansion, requiring a diagnostic test with high NPV. Interestingly, two possibilities exist to improve the NPV of the caval index. The first one is to choose a threshold value according to the clinical question with optimized NPV, rather than values with the highest Youden index. For example, the threshold value of 15% shows a NPV of 100% in Lanspa study and 82% in our study. The second way to improve the diagnostic accuracy of the caval index is to use a standardized inspiratory maneuver, which improves NPV without altering PPV. We found that the caval index assessed under standardized ventilation predicted fluid responsiveness with a NPV of 88%, and PPV of 86% at the threshold value of 39% (Table 1). Considering that the caval index is a noninvasive, easy to acquire parameter, showing a good positive predictive value, and a correct negative predictive value when optimized with a respiratory maneuver or an adapted threshold, we think that this index remains a very useful tool in clinical routine at patient’s bedside to guide fluid therapy.