Clinical Predictors of Fluid-Responsive AKI

Background: It is challenging to predict in clinical practice whether an episode of AKI may be reversible with volume expansion. No simple tool exists to aid in this decision-making process. Methods: We prospectively enrolled patients at Yale-New Haven Hospital with AKI, as defined by the Kidney Disease: Improving Global Outcomes creatinine criteria. We collected fluid administration data and standardized all volumes based on sodium concentration. The exposure was receiving at least 2L of normal saline (NS) equivalent in the 24h post-AKI. The outcome of interest was fluid-responsiveness–a plateau or decrease in serum creatinine within the 48h period beginning 24h post-AKI. We excluded patients meeting criteria for acute heart failure exacerbation and those who received ≥1L of NS equivalent in the 24h pre-AKI. We used inverse-probability weighting to estimate the interaction effects of receiving volume with 60 variables of interest collected in the 24h pre- or at the time of AKI on the outcome. Results: Of 3206 patients with AKI, 2313 met inclusion criteria, and 143 received the exposure. The interaction effects of select clinical variables are shown in Figure 1. Significant positive predictors of fluid-responsiveness in AKI patients receiving volume included higher temperature, higher urine chloride, and serum calcium >10.5 mg/dL. Significant negative predictors included higher respiratory rate, higher urine protein-creatinine ratio, higher modified Sequential Organ Failure Assessment (mSOFA) score, lactate dehydrogenase >280 U/L, and having serum uric acid checked. AKI in which a nephrologist suspected pre-renal etiology (excluding cardiorenal and hepatorenal syndromes) trended toward higher odds of fluid-responsiveness. Conclusion: Common clinical variables may be helpful in identifying who will benefit from volume challenge when the fluid-responsiveness of AKI is unclear. Funding: Other U.S. Government Support