Date of Award

January 2025

Document Type

Thesis

Degree Name

Medical Doctor (MD)

Department

Medicine

First Advisor

Francis P. Wilson

Abstract

CLINICAL PREDICTORS OF FLUID-RESPONSIVE ACUTE KIDNEY INJURY

Cindy Khanh Nguyen, Abinet M. Aklilu, Yu Yamamoto, Bashar A. Kadhim, Claudia Coronel-Moreno, Francis P. Wilson. Section of Nephrology, Department of Internal Medicine, Yale University, School of Medicine, New Haven, CT.

Specific Aims: We aimed to explore and identify common clinical variables that may be helpful in selecting patients who will benefit from volume challenge when the fluid-responsiveness of AKI is unclear.

Methods: Our study is a retrospective secondary analysis of a clinical trial that prospectively enrolled adult patients at Yale-New Haven Hospital with AKI defined by the 2012 Kidney Disease: Improving Global Outcomes Clinical Practice Guidelines for AKI serum creatinine criteria. We collected laboratory and fluid administration data from the electronic health record and standardized all fluid volumes based on sodium concentration. All clinical variables of interest were recorded in the 24 h prior to or at the time of AKI. We defined our exposure as receiving at least 2 liters (L) of normal-saline equivalent in the 24 hours (h) after AKI. Of 2,313 eligible study patients, 143 (6.2%) received the exposure. Our primary outcome was AKI improvement, defined by a plateau or decrease in serum creatinine within the 48-h period beginning 24 h after AKI. Our secondary outcomes were initiation of dialysis and death within 14 days after AKI. We created a propensity score to predict the probability of receiving intravenous fluids (IVF), then applied overlap weighting to balance covariates in the exposed and unexposed groups. We examined the interaction effects of receiving IVF with the clinical variables of interest on the outcome.

Results: In unweighted analysis, receiving ≥ 2 L IVF trended toward AKI worsening. After overlap weighting, this relationship was attenuated. Higher maximum temperature, higher minimum pH, and greater number of hyaline casts on urinalysis within the 24 hours prior to AKI predicted better fluid-responsiveness of AKI. Meanwhile, higher respiratory rate and having serum uric acid checked predicted worse fluid-responsiveness of AKI. Blood urea nitrogen-creatinine ratio > 20 was not a significant predictor of fluid-responsiveness of AKI.

Conclusions: We have shown that the clinical variables of temperature, pH, hyaline casts on urinalysis, respiratory rate, and having serum uric acid checked were available at the time of AKI diagnosis and were significant predictors of fluid-responsiveness of AKI. Our study was limited by its retrospective nature; small size of the exposure group; lack of data on clinical variables from physical exam, imaging, and intake/output; large percentage of missing data for many key variables of interest; and the multiple comparisons problem. There are potentially other important predictors of fluid-responsive AKI that we did not find, either due to the study limitations or because there are clinical variables that we have not yet explored. Our eventual goal with this research is to develop a multivariable model to predict fluid-responsiveness of AKI to aid in clinical decision-making.

Comments

This thesis is restricted to Yale network users only. It will be made publicly available on 05/14/2026

Download

Share

COinS