In the Literature
0.9% Saline or Balanced Crystalloid Fluids for Critically Ill
Patients: SPLIT Decision?
Commentary on Young P, Bailey M, Beasley R, et al. Effect of a buffered crystalloid solution vs saline on acute kidney injury among
patients in the intensive care unit: the SPLIT randomized clinical trial. JAMA. 2015;314(16):1701-1710.
T he origins of the salt solution for intravenous
(IV) uid resuscitation are attributed to the
Scottish physician William OShaughnessy, who in
1831 recommended the use of a dilute salt solution to
treat the profound hypovolemic shock of cholera.1 The
rst use in a patient followed shortly thereafter, when
Thomas A. Latta administered a warmed IV solution of
two drachams of muriate, two scruples of carbonate of
soda to sixty ounces of water2p257 as salvage therapy
to patients dying of cholera at the Leith Inrmary in
Scotland.1 Thereafter, countless solutions of various
compositions have been used across a wide range of
clinical settings. Ironically, none are well aligned with
the composition of the most omnipresent IV uid used
in medicine today, 0.9% (normal) saline solution.
Although 0.9% saline solution has a similar sodium
concentration to the aqueous phase of plasma (serum
sodium concentration, 154 mmol/L), it has a higher
relative chloride concentration and lower pH.3
Currently, there is little consensus on the ideal com-
position of IV uid and the optimal rate and volume for
administration.4 This has contributed in part to needless
variation in practice and potential for iatrogenic
harm.5-7 Recently, growing circumstantial evidence,
mostly from observational studies, suggests there are
signicant dose/rate-dependent adverse effects asso-
ciated with use of 0.9% saline solution for acute
resuscitation attributed largely to the chloride load,
including metabolic acidosis, hyperkalemia, acute
kidney injury (AKI), and kidney replacement ther-
apy.8-10 Although alternative buffered or balanced
solutions that more closely align with plasma are
widely available, no solution is considered awlessly
matched and there is limited evidence of comparative
benet or toxicity.11 Thus, despite medicines deeply
anchored condence in IV uid therapy, numerous
fundamental questions about the efcacy and safety of
its composition for acute resuscitation or as mainte-
nance therapy remain. Accordingly, the SPLIT (0.9%
Saline Versus Plasma-Lyte 148 for Intensive Care Unit
Fluid Therapy) trial is a timely contribution.12
WHAT DOES THIS IMPORTANT STUDY SHOW?
The SPLIT trial used a novel, double-blind, cluster-
randomized, crossover design.12 All patients admitted
to 1 of 4 academic intensive care units (ICUs) in New
Zealand between April and October 2014 who
received crystalloid uid therapy, not necessarily for
acute resuscitation, were eligible.
Am J Kidney Dis. 2016;68(1):11-14
Randomization occurred at the level of the ICU.
Each ICU was allocated to 4 alternating 7-week
blocks of study uids, such that each ICU crossed
over to provide each uid type twice. Frequency,
dose, and rate of uid prescription were at the
discretion of the treating clinician. The primary end
point was incidence of AKI.13 Box 1 details the
secondary and a priori planned subgroup analyses.14
Participant mean age was 60 years; two-thirds were
men, the burden of comorbid conditions was low (no
description of chronic kidney disease [CKD]), and
most were admitted following elective surgery (57%),
specically cardiovascular (49%). Mean APACHE II
(Acute Physiology and Chronic Health Evaluation II)
scores were 14.1, two-thirds received mechanical
ventilation, and mean baseline and admission serum
creatinine (Scr) values were 0.98 to 0.99 and 1.15 to
1.18 mg/dL, respectively. Prior to enrollment, two-
thirds had received a median volume of 1,000 to
1,200 mL of Plasma-Lyte 148 (PL-148; Baxter),
while the remaining one-third received a median
volume of 0 mL of 0.9% saline solution. The cumu-
lative dose of study uid given was small and similar
between those allocated to 0.9% saline solution
(median, 2,000 [interquartile range (IQR), 1,000-
3,500] mL) and PL-148 (2,000 [IQR, 1,000-3,250]
mL), with more than half being given in the rst day.
AKI occurred in 9.2% and 9.6% in those receiv-
ing 0.9% saline solution and PL-148 (P 5 0.77),
respectively. No data described the timing of AKI
relative to study uid exposure. Despite blinding,
66% of treating clinicians (n 5 36/55) correctly
guessed the uid type being administered. Kidney
replacement therapy was used infrequently and was
not different between groups (3.3% for PL-148 and
3.4% for 0.9% saline solution; P 5 0.91). There was
no signicant heterogeneity of treatment effect by
study uid across prespecied subgroups or differ-
ences in any secondary end point. However, there was
evidence of interaction between incidence of AKI and
Originally published online February 17, 2016.
Address correspondence to Sean M. Bagshaw, MD, MSc, Di-
vision of Critical Care Medicine, University of Alberta, 2-124E
Clinical Sciences Building, 8440-112 Street NW, Edmonton,
Alberta T6G 2B7, Canada. E-mail: bagshaw@ualberta.ca
2016 by the National Kidney Foundation, Inc.
0272-6386
http://dx.doi.org/10.1053/j.ajkd.2016.02.028
11

Box 1. Summary of SPLIT Trial Design
Design: Double-blind, cluster-randomized, double crossover
trial
Setting: 4 academic ICUs in New Zealand: 3 multidisci-
plinary ICUs and 1 specialized cardiothoracic and vascular
surgery ICU
Population: Inclusionall patients admitted to the study
ICUs who received crystalloid fluid therapy, as indicated by
local standards of practice. Exclusionpatients with end-
stage kidney disease receiving maintenance dialysis, those
expect to receive dialysis within 6 h of eligibility, and those
admitted to ICU for organ donation or end-of-life care
Intervention: 1,000-mL bags of Plasma-Lyte 148
Comparator: 1,000-mL bags of 0.9% saline solution
Outcomes: Primaryincidence of AKI, defined according to
Scr criteria of the RIFLE classification scheme. Secondary
peak relative change in Scr, receipt of kidney replacement
therapy, receipt and duration of mechanical ventilation, ICU
and hospital lengths of stay, ICU readmission, and all-cause
mortality through 90 d
Planned subgroup analyses of primary outcome and
hospital mortality: Sepsis, trauma with/without traumatic
brain injury, cardiac surgery, and admission APACHE II
score $ 25
Abbreviations: AKI, acute kidney injury; APACHE, Acute
Physiology and Chronic Health Evaluation; ICU, intensive care
unit; Scr, serum creatinine; SPLIT, 0.9% Saline Versus Plasma-
Lyte 148 for Intensive Care Unit Fluid Therapy.
study site. This was likely either due to chance or
attributable to variability in population baseline and
ICU-specic risks for AKI, though no further details
were provided.
The ndings of the SPLIT trial are certainly strength-
ened by its robust design, protocol adherence, and
transparent reporting. However, SPLIT has limita-
tions. First, SPLIT was a feasibility study intended
to interrogate a novel cluster crossover design.14
Second, SPLIT did not have an a priori sample
size estimation, such that when the size of the cohort
is considered in the context of the low event rate for
AKI, it is clearly inadequately powered to detect
relatively small but potentially important differences
in toxicity risk between uid types. Third, average
APACHE II scores were low, most patients were
postoperative, and use of kidney replacement therapy
was uncommon. These observations imply that the
study cohort was not enriched with patients at
increased risk for adverse kidney outcomes. Simi-
larly, the prevalence of CKD and heterogeneity in
treatment effect in this subgroup was not explored.
Fourth, SPLIT used a surrogate measure as a primary
end point, and the timing and frequency of Scr
measurements were not standardized. A total of 6.5%
had no Scr measured in the ICU. This likely
occurred among those with low perceived risk for
AKI, whereas in those at higher risk, more frequent
measurements were likely performed. This would put
SPLIT at risk for misclassication and ascertainment
12
Galm and Bagshaw
bias despite sensitivity analyses exploring extreme
case scenarios in missing data. Similarly, if the
toxicity of 0.9% saline solution is believed to be
attributed to the chloride load given to patients, data
for serum chloride concentration would further sup-
port biological plausibility; however, these data were
not reported. Finally, uid exposure occurring prior
to enrollment (predominantly PL-148), the small to-
tal dose of study uid given over a longer period,
and the lack of standardized administration all likely
dilute the capacity for SPLIT to detect any signal of
harm or benet associated with use of 0.9% saline
solution or PL-148.
HOW DOES THIS STUDY COMPARE WITH
PRIOR STUDIES?
Recently, several mostly observational studies have
described fewer complications and improved out-
comes among patients exposed to IV uids with lower
chloride concentrations.15 These data are consistent
with experimental and small clinical studies of
humans8,16,17 and the biological plausibility that 0.9%
saline solution is not physiologic due to the high
chloride concentration and lower relative strong ion
difference compared to plasma.18 Accordingly, rapid
and/or large volume administration, as in acute
resuscitation, can directly incite an iatrogenic hyper-
chloremic metabolic acidosis.8,19 This acidosis may
theoretically be exaggerated among susceptible pa-
tients, such as those with CKD, due to diminished
capacity to handle and excrete chloride.
The use of balanced crystalloid solutions in
diabetic ketoacidosis, despite the concern that the
added potassium content (potassium concentration,
5.0 mmol/L in PL-148) may exacerbate hyperkalemia,
was found in a small randomized trial to be associated
with greater increment in base decit and lower serum
chloride concentration compared to 0.9% saline solu-
tion.16 In a small randomized trial of intraoperative
0.9% saline solution compared to PL-148, 0.9% saline
solution contributed to greater postoperative metabolic
acidosis.19 In an observational study of adults under-
going major open abdominal surgery, the use of 0.9%
saline solution compared to PL-148 was associated
with greater electrolyte disturbances, more blood
transfusions and acidosis investigations, and higher
rates of complications, including receiving acute kid-
ney replacement therapy.9
The SPLIT trial adds to our body of knowledge on
this issue. SPLIT is the rst large-scale pragmatic
comparative effectiveness evaluation of crystalloid
uid composition in ICU patients. However, SPLIT
did not evaluate uids in the same context as many
of these aforementioned studies, in which the uid
doses given were larger and over shorter periods.
Moreover, SPLIT did not specically evaluate for a
Am J Kidney Dis. 2016;68(1):11-14
In the Literature
dose-response association between uid type and
hazard of adverse outcome. As an example, a recent
observational study of 0.9% saline solution use,
compared to balanced crystalloids, was associated
with higher risk for in-hospital death. However, there
was no signicant difference in AKI or kidney re-
placement therapy use in this large cohort of
nonsurgical patients with sepsis.20
Thus, while the ideal electrolyte solution for IV
uid therapy in acute resuscitation is undiscovered,
and may never be, SPLIT along with accumulated
data clearly reinforce the importance of baseline risk
and context when considering the potential hazard
associated with IV uids.
WHAT SHOULD CLINICIANS AND
RESEARCHERS DO?
The SPLIT trial was a feasibility and comparative
effectiveness trial, primarily aimed at detecting
whether there was clinically important toxicity and/or
harm associated with 0.9% saline solution when
compared to PL-148. The key inferences from SPLIT
could be either: (1) given it was a feasibility trial and
woefully underpowered, it should be quickly scaled to
provide more robust data suitable to inform clinical
practice; or (2) among low-acuity predominantly
postoperative ICU patients at relatively low risk for
AKI, a small total dose of 0.9% saline solution
compared to PL-148 (w2,000 mL), given over an
extended period (about 1-3 days), does not appear to
provoke an increased hazard for AKI. An alternative
perspective of these ndings would suggest that the
relatively healthy patients enrolled in SPLIT, who
received a small dose of uid, were more than able to
accommodate the hazard of 0.9% saline solution
compared to PL-148 (if truly present), whereas in an
enriched high-risk cohort, this may not have
occurred.21 This is analogous to the hazard of AKI,
kidney replacement therapy, and death associated
with use of hydroxyethyl starch among ICU patients
that appeared to be modied by case-mix and illness
severity.22,23
Importantly, although these data are reassuring for
lower risk patients receiving a little uid, they do not
provide clarity of the comparative effectiveness of
these uids for patients with greater baseline sus-
ceptibility to their toxic effects (ie, CKD) or those
who may receive signicantly more rapid and larger
amounts. The SPLIT investigators acknowledge this
and rightfully conclude by stating that additional
high-quality trials are needed.12 Future clinical trials
should also integrate a careful evaluation of the
impact of the dose-response hazard associated with IV
uids on outcomes.21
The SPLIT trial should remind all clinicians to
be respectful to the potential iatrogenic harm, both
Am J Kidney Dis. 2016;68(1):11-14
qualitative and quantitative, associated with uid
therapy. Similar to any drug used in acutely ill hos-
pitalized patients, the prescription of IV uid therapy
should be context specic, with the dose, rate, and
end points for administration clearly specied, such
that it is given to the right patient, at the right time,
and in the right context.
Brandon Galm, MD
Sean M. Bagshaw, MD, MSc
University of Alberta
Edmonton, Canada
ACKNOWLEDGEMENTS
Support: Dr Bagshaw is supported by a Canada Research Chair
in Critical Care Nephrology.
Financial Disclosure: Dr Bagshaw has received research sup-
port from and consulted for Baxter. Dr Galm declares that he has
no relevant nancial interests.
Peer Review: Evaluated by the Deputy Editor and the Editor-in-
Chief.
REFERENCES
1. Awad S, Allison SP, Lobo DN. The history of 0.9% saline.
Clin Nutr. 2008;27(2):179-188.
2. Lewins R. Injection of Saline Solutions into the Veins.
London Medical Gazette. May 26, 1832:257.
3. Moritz ML, Ayus JC. Maintenance intravenous uids in
acutely ill patients. N Engl J Med. 2015;373(14):1350-1360.
4. Kellum JA, Mythen MG, Shaw AD. The 12th Consensus
Conference of the Acute Dialysis Quality Initiative (ADQI XII).
Br J Anaesth. 2014;113(5):729-731.
5. Finfer S, Liu B, Taylor C, et al. Resuscitation uid use in
critically ill adults: an international cross-sectional study in 391
intensive care units. Crit Care. 2010;14(5):R185.
6. National Institute for Health and Care Excellence. NICE
guidelines [CG174]: intravenous uid therapy in over 16s in hos-
pital: NICE; 2013 [cited 2015]. https://www.nice.org.uk/Guidance/
CG174. Accessed November 10, 2015.
7. Arieff AI, Ayus JC, Fraser CL. Hyponatraemia and death or
permanent brain damage in healthy children. BMJ. 1992;304(6836):
1218-1222.
8. Chowdhury AH, Cox EF, Francis ST, Lobo DN.
A randomized, controlled, double-blind crossover study on the
effects of 2-L infusions of 0.9% saline and Plasma-Lyte(R) 148 on
renal blood ow velocity and renal cortical tissue perfusion in
healthy volunteers. Ann Surg. 2012;256(1):18-24.
9. Shaw AD, Bagshaw SM, Goldstein SL, et al. Major compli-
cations, mortality, and resource utilization after open abdominal sur-
gery: 0.9% saline compared to Plasma-Lyte. Ann Surg. 2012;255(5):
821-829.
10. Yunos NM, Bellomo R, Bailey M. Chloride-restrictive
uid administration and incidence of acute kidney injuryreply.
JAMA. 2013;309(6):543-544.
11. Young JB, Utter GH, Schermer CR, et al. Saline versus
Plasma-Lyte A in initial resuscitation of trauma patients: a ran-
domized trial. Ann Surg. 2014;259(2):255-262.
12. Young P, Bailey M, Beasley R, et al. Effect of a buffered
crystalloid solution vs saline on acute kidney injury among pa-
tients in the intensive care unit: the SPLIT randomized clinical
trial. JAMA. 2015;314(16):1701-1710.
13

13. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P;
Acute Dialysis Quality Initiative workgroup. Acute renal failure -
denition, outcome measures, animal models, uid therapy and
information technology needs: the Second International Consensus
Conference of the Acute Dialysis Quality Initiative (ADQI)
Group. Crit Care. 2004;8(4):R204-R212.
14. Reddy SK, Bailey MJ, Beasley RW, et al. A protocol for
the 0.9% Saline Versus Plasma-Lyte 148 for Intensive Care Fluid
Therapy (SPLIT) study. Crit Care Resusc. 2014;16(4):274-279.
15. Krajewski ML, Raghunathan K, Paluszkiewicz SM,
Schermer CR, Shaw AD. Meta-analysis of high- versus low-
chloride content in perioperative and critical care uid resuscita-
tion. Br J Surg. 2015;102(1):24-36.
16. Mahler SA, Contrad SA, Wan H, Arnold TC. Resuscitation
with balanced electrolyte solution prevents hyperchloremic meta-
bolic acidosis in patients with diabetic ketoacidosis. Am J Emerg
Med. 2011;29:670-674.
17. Wilcox CS. Regulation of renal blood ow by plasma
chloride. J Clin Invest. 1983;71(3):726-735.
14
Galm and Bagshaw
18. Wakim KG. Normal 0.9 per cent salt solution is
neither normal nor physiological. JAMA. 1970;214(9):
1710.
19. McFarlane C, Lee A. A comparison of Plasmalyte 148 and
0.9% saline for intra-operative uid replacement. Anaesthesia.
1994;49(9):779-781.
20. Raghunathan K, Shaw A, Nathanson B, et al. Association
between the choice of IV crystalloid and in-hospital mortality
among critically ill adults with sepsis*. Crit Care Med.
2014;42(7):1585-1591.
21. Kellum JA, Shaw AD. Assessing toxicity of intravenous
crystalloids in critically ill patients. JAMA. 2015;314(16):1695-
1697.
22. Myburgh JA, Finfer S, Bellomo R, et al. Hydroxyethyl
starch or saline for uid resuscitation in intensive care. N Engl J
Med. 2012;367(20):1901-1911.
23. Perner A, Haase N, Guttormsen AB, et al. Hydroxyethyl
starch 130/0.42 versus Ringers acetate in severe sepsis. N Engl J
Med. 2012;367(2):124-134.
Am J Kidney Dis. 2016;68(1):11-14