The Study and Background
The Study
Association Between a Chloride-Liberal vs Chloride-Restrictive Intravenous Fluid Administration Strategy and Kidney Injury in Critically Ill Adults
Yunos NM, Bellomo R, Hegarty C, Story D, Ho L, Bailey M
JAMA. 2012;308:1566-1572
Background
Like many women, I want to apologize for my behavior to all the healthcare professionals who helped me to deliver my children. But my apology is for a unique reason. As my seventh IV fluid bag was being hung, I tried to explain the thirst mechanism, what my antidiuretic hormone level must have been, and why I would not follow the NPO status for water. I also remember saying to the anesthesiologist as I saw his choice of lactated Ringer's, "Oh good. I was feeling quite lactate deplete."
Turns out, he may very well have been right.
A recent study in JAMA compared outcomes of patients in a single-center intensive care unit (ICU) during 2 periods. For each period, the ICU adopted a different intravenous fluid as the default of choice. During the first period, the default fluid was "normal saline" (0.9% with 150 mmol/L of chloride), referred to as a chloride-liberal solution. During the second period, 3 default solutions considered to be chloride-restrictive were available: a lactated crystalloid solution (109 mmol/L chloride), a balanced buffered solution (98 mmol/L chloride), and a 20% albumin solution (19 mmol/L chloride).
This study was prospectively designed to examine the rate of prespecified outcomes during these 2 periods and to draw conclusions based on causality with respect to the impact of chloride-liberal vs chloride-restrictive IV fluids. Because the constitutions of all of these intravenous fluids were in routine use, the study was considered to be only a change in ICU policy, and the institutional internal review board approved it without requiring individual patient consent. During each period, the specific IV fluids were available to clinicians on the basis of individual needs of the patients. When a fluid constitution was not specifically indicated, the default (chloride-liberal for the first period and chloride-restrictive for the second period) was used.
The number and characteristics of patients on entry into the ICU during the 2 periods were similar. Of note, several prognostic scoring systems used to compare the 2 cohorts at baseline, including APACHE II, APACHE III, and SAPS, were all without significant differences (P = .63, .99, and .94, respectively). Serum creatinine on admission was 90 µmol/L (69-125 µmol/L) and 86 µmol/L (67-121 µmol/L) (P = .07), respectively, for the chloride-liberal and chloride-restrictive cohorts on admission to the ICU. Patients were followed prospectively for the occurrence of acute kidney injury (AKI) based on the RIFLE criteria.
More patients in the chloride-liberal cohort vs the chloride-restrictive cohort developed AKI using any definition of the RIFLE criteria (37% vs 23%, respectively) and had the most severe category of injury and failure (14% vs 8.4%; P < .001). Additionally, the proportion of patients requiring dialysis was also greater in the chloride-liberal cohort: 10% (8.1%-12%) compared with 6.3% (4.6%-8.1%; P < .005). After adjusting for sex, APACHE III score, diagnosis, operative status, baseline serum creatinine level, and admission type (elective or emergency), the differences in rates of AKI and use of renal replacement therapy remained similar and significant between the chloride-liberal and chloride-restrictive cohorts (P = .001 and .004, respectively).
Although the rate of AKI was different between cohorts, no difference in mortality was seen. Nine percent (7%-11%) of patients in the chloride-liberal group experienced intra-ICU mortality vs 8% (6%-10%) in the chloride-restrictive group (P = .42). This study design has limitations compared with a traditional randomized trial, but it is important to note that its findings are very consistent with previous observations showing a benefit with a chloride-restrictive strategy. Taken together, they suggest that chloride-restricted fluids are associated with a mechanism of less renal vasoconstriction than are chloride-liberal fluids.