Balanced crystalloid versus saline in critically ill adults

In critically ill patients, the use of a balanced crystalloid rather than saline may lead to a reduction in the composite outcome measure, MAKE-30. Level of evidence: 2b (randomised controlled trial (RCT) with a high risk of bias)

Appraised by: Matthew JW Rowe and Chris Cairns

Citation: Semler MW, Self WH, Wanderer JP, et al.; for the SMART investigators and the Pragmatic Critical Care Research Group. Balanced crystalloid versus saline in critically ill adults. N Engl J Med 2018; 378: 829–839.

Lead author: Matthew W Semler; matthew.w.semler@vanderbilt.edu

Three-part clinical question:

Patients: All adult patients admitted to one of five intensive care units (ICUs) at a single academic centre in the United States.

Intervention: Each of the five ICUs were assigned to use either balanced crystalloid solution (either lactated Ringer’s solution or Plasma-lyte A) or 0.9% saline for each month of participation in the trial. All adult patients being admitted to each of the ICUs, therefore received whichever intravenous fluid that particular unit had been assigned. ICUs were randomised to use saline during even numbered months and balanced crystalloid during odd numbered months or vice versa.

Outcomes: The primary outcome was the proportion of patients who met one or more criteria for a Major Adverse Kidney Event at 30 days (MAKE30): Mortality, new receipt of renal replacement therapy (RRT) or persistent renal dysfunction (defined as a final inpatient creatinine value ≥200% of the baseline value). Censored at hospital discharge or 30 days post-enrolment, whichever occurred first. Secondary outcomes were split into clinical and renal outcomes. Secondary clinical outcomes included in-hospital death before ICU discharge or at 30 days or 60 days, ICU-free days, ventilator-free days, vasopressor-free days and days alive and free of RRT during the 28 days after enrolment. Secondary renal outcomes included new receipt of RRT, persistent renal dysfunction and the development of a stage 2 or higher acute kidney injury (AKI).

Study design:

Pragmatic, single centre, un-blinded, cluster-randomised, multiple cross-over trial with intention-to-treat analysis.

Study patients:

Inclusion criteria: All adult patients admitted to one of five ICUs (medical, neurological, cardiac, trauma and surgical) at a single academic centre in the U.S. (Vanderbilt Medical Centre, Nashville, Tennessee) over a 22-month period.

Exclusion criteria: Patients aged under 18 years of age.

Study groups: ICUs admitting primarily from the emergency department (ED) were allocated the same fluid at the same time to ensure fluid administered in the ED was likely to be the same as that given in the ICU. In contrast, ICUs admitting patients from theatre areas were assigned to the alternative crystalloid solution. Fluids administered in theatres were then matched with those allocated to the post-surgical ICU for that particular month. In total, 15,802 patients were assigned to receive either balanced crystalloid or saline; 7942 patients received balanced crystalloid whilst 7860 patients received saline. A sample size of 14,000 patients was adjusted from an initial number of 8000 patients following analysis of observational data, obtained during the preceding year, indicating the incidence of the outcome in the saline group would be 15%.¹ This would provide a power of 90% to detect and absolute difference of 1.9 percentage points between the two groups with a type-1 error rate of 0.05.

Results:

The published results would tend to suggest that the use of balanced crystalloid solution, as opposed to saline, in critically ill patients resulted in an absolute difference of 1.1 percentage points in favour of using balanced crystalloid in the primary outcome (see table 1).

Table 1.

Published results.

Outcome	ER-balanced crystalloid (n = 7942)	ER 0.9% saline (n = 7860)	MOR	COR	NNT
MAKE-30	0.143	0.154	0.91	0.9	94
(95% CI)			(0.84–0.99)	(0.82–0.99)

COR: conditional odds ratio; ER: event rate; MAKE-30: Major Adverse Kidney Event at 30 days; MOR: marginal odds ratio; NNT: number needed to treat in order to avoid one MAKE-30 outcome; 95% CI: 95% confidence interval.

However, these results were generated using adjusted variables. Using CATmaker™ software to analyse the published raw results suggests that the 95% CIs for RRR, ARR and therefore NNT include zero indicating a non-significant statistical result (see table 2). CATmaker™ analysis of the individual components of the composite outcome confirms that none of them reach statistical significance independently (Table 3).

Table 2.

CATmaker™ generated results.

Outcome	Time to outcome	CER	EER	RRR	ARR	NNT
Composite: death, new RRT or final creatinine >200% baseline	30 Days	0.154	0.143	7%	1.1%	ns
	95% Confidence intervals:			0% to 14%	0.0% to 2.2%

ARR: absolute risk reduction; CER: control event rate (saline); EER: experimental event rate (balanced solution); NNT: number needed to treat; RRR: relative risk reduction.

Table 3.

CATmaker™ generated results.

Outcome	Time to outcome	CER	EER	RRR	ARR	NNT
In-hospital death	30 Days	0.111	0.103	7%	0.8%	ns
	95% Confidence intervals:			−1% to 16%	−0.2% to 1.8%
New RRT	30 Days	0.029	0.025	14%	0.4%	ns
	95% Confidence intervals:			−4% to 31%	−0.1% to 0.9%
Creatinine >200% baseline	30 Days	0.066	0.064	3%	0.2%	ns
	95% Confidence intervals:			−9% to 15%	−0.6% to 1.0%

ARR: absolute risk reduction; CER: control event rate (saline); EER: experimental event rate (balanced solution); NNT: number needed to treat; RRR: relative risk reduction.

For secondary outcomes, there was no significant difference in the number of ICU-free days, ventilator free days or the incidence of stage 2 or higher AKI. There was a significantly lower incidence of hyperchloraemia (chloride >110 mmol/l) or low serum bicarbonate (<20 mmol/l) in the balanced group.

EBM questions:

1. Do the methods allow accurate testing of the hypothesis? No. Although this was a large, pragmatic, randomised trial with a large enough sample size to generate sufficient power to detect small differences in clinical outcome, there was a high risk of bias due to the lack of blinding. The use of a composite primary outcome (MAKE-30) resulted in a positive result that was only just statistically significant using corrected data but failed to reach significance using raw data. Furthermore, the clinical significance of this composite is unclear. None of the individual components of the composite demonstrated a statistical difference.

2. Do the statistical tests correctly test the results to allow differentiation of statistically significant results? See question 1. Sample size and power calculations seem appropriate.

3. Are the conclusions valid in light of the results? Perhaps. The study authors concluded that, in critically ill adults, the use of balanced crystalloid over saline resulted in an absolute difference of 1.1 percentage points in favour of balanced crystalloid in reducing the incidence of a MAKE-30 outcome. This is only the case if adjusted data is used. The methodology did not allow any significant difference in the effects of Ringers or Plasma-Lyte to be identified.

4. Did results get omitted and why? No. No omitted results were declared; however, treating clinicians had the power to alter which fluid a patient received if they felt that it may exacerbate the patient’s condition. These patients were analyses in an intention-to-treat fashion.

5. Did the authors suggest any further areas of research? No.

6. Did they make any recommendations based on the results and are they appropriate? No.

7. Is the study relevant to my clinical practice? Perhaps. The patient mix is not typical of all UK ICUs. It is very much a mixed ICU/HDU patient cohort (34% ventilated, 26% vasopressors, 10% predicted in-hospital mortality). This needs taking into account when assessing applicability. It is unfortunate that Ringers lactate and Plasma-lyte A were grouped together making an informed choice between these two fluids impossible.

8. What level of evidence does this study represent? 2B. A randomised control trial with a high risk of bias.

9. What grade of recommendation can I make on this result alone? None.

10. What grade of recommendation can I make when this study is considered along with other available evidence? None. More studies in the critical care setting are required.

11. Should I change my practice because of these results? No. We should await the results of the upcoming “plasma-lyte 148 versus saline (PLUS) study.² This is a multi-centre, blinded, randomised control trial with a primary outcome of 90-day mortality

12. Should I audit my practice because of these results? No. There are no specific indications to audit practice based on the findings of this study.

Appraised by:

Matthew JW Rowe, Core Trainee 3, Great Western Hospitals NHS Trust, Swindon, UK

Chris Cairns, Consultant, Intensive Care, Forth Valley Royal Hospital