Practical application of the Stewart acid-base model in Surgical Intensive Care Unit patients undergoing intravenous diuresis

Document Type

Conference Proceeding

Publication Date



INTRODUCTION: Acid-base disturbances in the ICU have traditionally been analyzed using a qualitative bicarbonate-based approach that has been widely criticized. A more contemporary, quantitative approach, the Stewart model, provides a more physiochemical, clinically relevant method to interpret acid-base disturbances. Continuous infusion (CI) loop diuretic therapy is often utilized to manage fluid overload and weaning from mechanical ventilation in the ICU, but may cause an increase in Strong Ion Difference (SID) and metabolic alkalosis. There is currently no data characterizing the effects of loop diuretics using the Stewart model, which may reveal practical management options to prevent or delay the development of metabolic alkalosis.

RESEARCH QUESTION OR HYPOTHESIS: Based on Stewart model calculations, we sought to test the strength of correlation between predicted and observed systemic acid-based status during CI loop diuretic therapy in the ICU.

STUDY DESIGN: A prospective, single-center, observational study conducted in the Surgical ICU of a large academic medical center.

METHODS: Ten critically ill patients who received CI furosemide were included. Over a 72-hour period, intake and output volumes, electrolyte content of all fluids administered, plasma and urine electrolytes, urine pH, and venous blood gases were collected. The predicted (calculated using volume status, SID of fluids in, SID of urine out, and baseline SID) and observed change in acid-based status was compared for each day using Spearmans Correlation Coefficient. Statistical analyses were conducted using SPSS Statistics (v24, SPSS Inc.)

RESULTS: At day 1 the mean observed plasma SID was 47.49(3.46) mEq/L and the predicted SID value was 49.54 (5.6) mEq/L. Day 1 observed plasma SID was positively correlated with thepredicted SID value (r=.080, p=0.01). Day 2 and 3 correlations of observed and predicted SID were not statistically significant.

CONCLUSION: Using the Stewart model, the expected increasein SID and metabolic alkalosis was able to be predicted for agroup of ICU patients receiving CI furosemide.




Presented at the American College of Clinical Pharmacy Annual Meeting in Phoenix, Arizona, October 7-10, 2017.

Abstract published in Pharmacotherapy, Volume 37, Issue 12, December 2017, Page e149, Abstract #86: https://doi.org/10.1002/phar.2052