HHS Public Access

Author manuscript
N Engl J Med. Author manuscript; available in PMC 2018 September 01.
Author Manuscript

Published in final edited form as:

N Engl J Med. 2018 March 01; 378(9): 829–839. doi:10.1056/NEJMoa1711584.

Balanced Crystalloids versus Saline in Critically Ill Adults

Matthew W. Semler, M.D., Wesley H. Self, M.D., M.P.H., Jonathan P. Wanderer, M.D., Jesse
M. Ehrenfeld, M.D., M.P.H., Li Wang, M.S., Daniel W. Byrne, M.S., Joanna L. Stollings,
Pharm.D., Avinash B. Kumar, M.D., Christopher G. Hughes, M.D., Antonio Hernandez, M.D.,
Oscar D. Guillamondegui, M.D., M.P.H., Addison K. May, M.D., Liza Weavind, M.B., B.Ch,
Jonathan D. Casey, M.D., Edward D. Siew, M.D., Andrew D. Shaw, M.B., Gordon R. Bernard,
M.D., and Todd W. Rice, M.D. for the SMART Investigators and the Pragmatic Critical Care
Author Manuscript

Research Group*
Department of Medicine, Division of Allergy, Pulmonary, and Critical Care Medicine (M.W.S.,
J.D.C., G.R.B., T.W.R.), the Departments of Emergency Medicine (W.H.S.), Anesthesiology
(J.P.W., J.M.E., A.B.K., C.G.H., A.H., L. Weavind, A.D.S.), Biomedical Informatics (J.P.W., J.M.E.),
Surgery (J.M.E., O.D.G., A.K.M.), Health Policy (J.M.E.), Biostatistics (L. Wang, D.W.B.), and
Pharmaceutical Services (J.L.S.), and the Division of Nephrology and Hypertension, Vanderbilt
Center for Kidney Disease and Integrated Program for Acute Kidney Disease (E.D.S.) — all at
Vanderbilt University Medical Center, Nashville

Abstract
BACKGROUND—Both balanced crystalloids and saline are used for intravenous fluid
Author Manuscript

administration in critically ill adults, but it is not known which results in better clinical outcomes.

METHODS—In a pragmatic, cluster-randomized, multiple-crossover trial conducted in five

intensive care units at an academic center, we assigned 15,802 adults to receive saline (0.9%
sodium chloride) or balanced crystalloids (lactated Ringer’s solution or Plasma-Lyte A) according
to the randomization of the unit to which they were admitted. The primary outcome was a major
adverse kidney event within 30 days — a composite of death from any cause, new renal-
replacement therapy, or persistent renal dysfunction (defined as an elevation of the creatinine level
to ≥200% of baseline) — all censored at hospital discharge or 30 days, whichever occurred first.

RESULTS—Among the 7942 patients in the balanced-crystalloids group, 1139 (14.3%) had a
major adverse kidney event, as compared with 1211 of 7860 patients (15.4%) in the saline group
(marginal odds ratio, 0.91; 95% confidence interval [CI], 0.84 to 0.99; conditional odds ratio, 0.90;
95% CI, 0.82 to 0.99; P = 0.04). In-hospital mortality at 30 days was 10.3% in the balanced-
Author Manuscript

crystalloids group and 11.1% in the saline group (P = 0.06). The incidence of new renal-
replacement therapy was 2.5% and 2.9%, respectively (P = 0.08), and the incidence of persistent
renal dysfunction was 6.4% and 6.6%, respectively (P = 0.60).

Address reprint requests to Dr. Rice at the Department of Medicine, Vanderbilt University Medical Center, T-1218 MCN, 1161 21st
Ave. S., Nashville, TN 37232, or at [email protected].
*A complete list of the SMART Investigators is provided in the Supplementary Appendix, available at NEJM.org.
Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.
 Semler et al. Page 2

CONCLUSIONS—Among critically ill adults, the use of balanced crystalloids for intravenous
Author Manuscript

fluid administration resulted in a lower rate of the composite outcome of death from any cause,
new renal-replacement therapy, or persistent renal dysfunction than the use of saline. (Funded by
the Vanderbilt Institute for Clinical and Translational Research and others; SMART-MED and
SMART-SURG ClinicalTrials.gov numbers, NCT02444988 and NCT02547779.)

Intravenous crystalloid solutions are commonly administered in critical care, yet the
question of whether crystalloid composition affects patient outcomes remains unanswered.1
Historically, 0.9% sodium chloride (saline) has been the most commonly administered
intravenous fluid.2,3 Data suggest that intravenous saline may be associated with
hyperchloremic metabolic acidosis,4 acute kidney injury,5 and death.6,7 Crystalloid solutions
with electrolyte compositions closer to that of plasma (balanced crystalloids, such as lactated
Ringer’s solution or Plasma-Lyte A) represent an increasingly used alternative to saline.8
Several observational studies6,9,10 and a before-and-after trial5 suggested that the use of
Author Manuscript

balanced crystalloids is associated with lower rates of acute kidney injury, renal-replacement
therapy, and death. However, in two pilot trials,11,12 no significant difference in any patient
outcome was reported between those who received balanced crystalloids and those who
received saline.

To determine the effect of isotonic crystalloid composition on clinical outcomes in critically

ill adults, we conducted the Isotonic Solutions and Major Adverse Renal Events Trial
(SMART), which compared the use of balanced crystalloids with the use of saline in patients
in medical (SMART-MED) and nonmedical (SMART-SURG) intensive care units (ICUs).
We hypothesized that the use of balanced crystalloids would result in a lower overall
incidence of death, new renal-replacement therapy, and persistent renal dysfunction than
saline.
Author Manuscript

METHODS
TRIAL DESIGN AND OVERSIGHT
We conducted a pragmatic, unblinded, cluster-randomized, multiple-crossover trial in which
the use of balanced crystalloids was compared with saline for intravenous fluid
administration among critically ill adults admitted to five ICUs at Vanderbilt University
Medical Center between June 1, 2015, and April 30, 2017. The trial was approved by the
institutional review board at Vanderbilt University with a waiver of informed consent (see
the Supplementary Appendix, available with the full text of this article at NEJM.org), was
registered online before initiation, and was overseen by an independent data and safety
monitoring board. The protocol, available at NEJM.org, and the statistical analysis plan were
published before the conclusion of enrollment.13 All authors vouch for the accuracy and
Author Manuscript

completeness of the data and for the fidelity of the trial to the protocol.

TRIAL SITES AND PATIENT POPULATION

All adults (18 years of age or older) who were admitted to a participating ICU during the
trial period were enrolled at the time of ICU admission (site characteristics are described in
the Supplementary Appendix). Enrolled patients who were discharged from the hospital
were eligible to participate again if they were readmitted to a participating ICU. We assessed

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 3

the effect of repeat hospitalizations in individual patients in sensitivity analyses. Patients

Author Manuscript

who were admitted to a non-ICU ward from the emergency department were enrolled in a
separate trial (Saline against Lactated Ringer’s or Plasma-Lyte in the Emergency
Department [SALT-ED]) in which balanced crystalloids and saline were compared among
adults who were not critically ill. The results of that trial are also reported in this issue of the
Journal.14

RANDOMIZATION
For each month of the trial, participating ICUs were assigned to use either balanced
crystalloids or saline for any intravenous administration of isotonic crystalloid. ICUs were
randomly assigned to use saline during even-numbered months and balanced crystalloids
during odd-numbered months, or vice versa (Fig. S1 in the Supplementary Appendix). To
allow coordination of crystalloid use between ICUs and the emergency department and
Author Manuscript

operating rooms, the three ICUs that admit the majority of patients from the emergency
department underwent randomization together, as did the two ICUs that admit the majority
of patients from operating rooms.13 Patients, clinicians, and investigators were aware of
group assignments.

TREATMENTS
Patients in the saline group received 0.9% sodium chloride when intravenous isotonic
crystalloid was administered, whereas patients in the balanced-crystalloids group received
either lactated Ringer’s solution or Plasma-Lyte A, according to the preference of the
treating clinician (Table S1 in the Supplementary Appendix). An electronic advisor within
the electronic order-entry system informed providers about the trial, asked about relative
contraindications to the assigned crystalloid, and, if none were present, guided providers to
Author Manuscript

order the assigned crystalloid. Relative contra-indications to the use of balanced crystalloids
included hyperkalemia and brain injury. The treating clinician determined the severity of
hyperkalemia or brain injury at which saline rather than balanced crystalloids would be used.
The unassigned crystalloid was also available from the pharmacy when clinicians believed it
to be required for the safe treatment of any patient.

The trial was coordinated with the emergency department and operating rooms so that when
feasible, patients being admitted to a participating ICU or receiving a surgical intervention
during ICU admission would receive the crystalloid assigned to that ICU.15 The need for
access to an intravenous crystalloid at all times precluded the use of washout periods, and
patients who remained in the ICU from the end of one calendar month to the start of another
may have been exposed to both types of crystalloid. The effect of dual exposure was
evaluated in prespecified sensitivity analyses.
Author Manuscript

DATA COLLECTION
We used data collected in routine care and electronically extracted from electronic health
records.12,16 These data included information on pre-enrollment renal function,
demographic characteristics, diagnoses, predicted risk of inhospital death, orders for
intravenous fluids and blood products, plasma electrolyte and creatinine values, receipt of
renal-replacement therapy, and vital status at hospital discharge. Trial personnel who were

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 4

unaware of group assignment performed manual chart reviews to confirm receipt of renal-
Author Manuscript

replacement therapy and identify indications for new renal-replacement therapy.

OUTCOMES
The primary outcome was the proportion of patients who met one or more criteria for a
major adverse kidney event within 30 days16–20 — the composite of death, new receipt of
renal-replacement therapy, or persistent renal dysfunction (defined as a final inpatient
creatinine value ≥200% of the baseline value) — all censored at hospital discharge or 30
days after enrollment, whichever came first. The National Institute of Diabetes and Digestive
and Kidney Diseases work group on clinical trials in acute kidney injury recommends the
use of a major adverse kidney event within 30 days as a patient-centered outcome for phase
3 trials.16,18 We determined a value for baseline creatinine level using a previously described
hierarchical approach in which creatinine values obtained during the year before
Author Manuscript

hospitalization were given priority over in-hospital measurements obtained before ICU
admission. The baseline creatinine level was estimated with a previously described three-
variable formula when no pre-enrollment measurements were available (for details, see the
Supplementary Appendix).16,21 Patients who had received renal-replacement therapy before
enrollment were ineligible to meet the criteria for new renal- replacement therapy or
persistent renal dysfunction but could qualify for the primary outcome if they died in the
hospital.

Secondary clinical outcomes included in-hospital death before ICU discharge or at 30 days
or 60 days, as well as ICU-free days, ventilator-free days, vasopressor-free days, and days
alive and free of renal-replacement therapy during the 28 days after enrollment.13 Secondary
renal outcomes included new receipt of renal-replacement therapy, persistent renal
dysfunction, acute kidney injury of stage 2 or higher as defined in the Kidney Disease:
Author Manuscript

Improving Global Outcomes criteria for creatinine level,22 the highest creatinine level during
the hospital stay, the change from baseline to the highest creatinine level, and the final
creatinine level before hospital discharge.13

STATISTICAL ANALYSIS
Complete details regarding the sample-size justification have been reported previously.13
Initially, we planned to enroll 8000 patients during 60 unit-months (12 months in five ICUs)
to detect a 12% relative between-group difference11,12 in the primary outcome of a major
adverse kidney event within 30 days, assuming a 22.0% incidence of the outcome in the
saline group on the basis of the findings in a previous report.19 We subsequently obtained
observational data for patients admitted to the ICUs involved in the trial in the year before
the trial began. These data suggested that the incidence of the outcome in the saline group
Author Manuscript

would be approximately 15.0%. To retain adequate power to detect the targeted difference in
relative risk, in collaboration with the data and safety monitoring board, the duration of the
trial was increased to 82 unit-months. Enrolling approximately 14,000 patients during 82
unit-months would provide power of 90% at a type I error rate of 0.05 to detect a relative
difference of 12% (an absolute difference of 1.9 percentage points) in the primary outcome
between groups.13 The data and safety monitoring board conducted two interim analyses;
details are provided in the Supplementary Appendix.

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 5

Analyses were conducted at the level of each patient’s hospitalization in an intention-to-treat

Author Manuscript

fashion. Continuous variables are reported as means and standard deviations or as medians
and interquartile ranges; categorical variables are reported as frequencies and proportions.

The primary analysis compared the incidence of the primary outcome in the balanced-
crystalloids and saline groups with a generalized, linear, mixed-effects model that included
fixed effects (group assignment, age, sex, race, source of admission, mechanical-ventilation
status, vasopressor receipt, diagnosis of sepsis, and diagnosis of traumatic brain injury) and
random effects (ICU to which the patient was admitted) (for details, see the Supplementary
Appendix).23,24 Both conditional (ICU-level) and marginal (population-level) effects are
reported.

Prespecified secondary analyses involved a similar approach. First, we compared secondary

outcomes between trial groups. Second, we performed subgroup analyses according to type
Author Manuscript

of ICU, source of admission, receipt of mechanical ventilation, receipt of vasopressors,

diagnosis of sepsis or traumatic brain injury (for details, see the Supplementary Appendix),
baseline renal function, predicted in-hospital mortality, and total volume of isotonic
crystalloid administered through day 30. Third, we conducted sensitivity analyses using
alternative approaches to addressing the issue of missing data on baseline creatinine level
(for details, see the Supplementary Appendix). Fourth, we performed sensitivity analyses
according to the volume of crystalloid administered, accounting for crossover and limiting
the analyses to each patient’s first ICU admission.13 Other between-group comparisons were
made with the Mann–Whitney rank-sum test for continuous variables and the chi-square test
for categorical variables.

A two-sided P value of less than 0.048 indicated statistical significance for the primary
Author Manuscript

outcome after accounting for interim analyses. All other analyses were considered to be
hypothesis-generating.13 With 14 secondary outcomes, the likelihood of observing a P value
of less than 0.05 for at least one secondary outcome by chance alone was 51.2%. All
analyses were performed with the statistical software R, version 3.3.0, with a prespecified
analysis code published before the conclusion of enrollment.13

RESULTS
BASELINE CHARACTERISTICS
In all, 15,802 patients from five ICUs were enrolled in the trial (Fig. S2 in the
Supplementary Appendix). The median age was 58 years, and 57.6% of patients were men.
More than one third of patients were receiving mechanical ventilation and one quarter were
receiving vasopressors at enrollment. There were no significant differences in baseline
Author Manuscript

characteristics between the patients assigned to receive balanced crystalloids (7942 patients)
and those assigned to receive saline (7860 patients) (Table 1, and Tables S2 and S3 in the
Supplementary Appendix).

FLUID THERAPY AND ELECTROLYTES

Because the fluid therapy provided in the emergency department and operating room was
coordinated with that provided in the ICU to which patients were being admitted, the

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 6

majority of pre-ICU fluid that patients received was consistent with trial-group assignment
Author Manuscript

(Table S4 in the Supplementary Appendix). The median volume of balanced crystalloids

administered to patients in the balanced-crystalloids group between ICU admission and
hospital discharge or 30 days (whichever occurred first) was 1000 ml (inter-quartile range, 0
to 3210), and the median volume of 0.9% sodium chloride administered to patients in the
saline group was 1020 ml (inter-quartile range, 0 to 3500) (Fig. 1, and Tables S5 and S6 in
the Supplementary Appendix). Only 426 patients (5.4%) in the balanced-crystalloids group
and 343 patients (4.4%) in the saline group received any volume of unassigned crystalloid as
a result of remaining in the ICU from one calendar month to the next (Table S5 in the
Supplementary Appendix). There was no significant between-group difference in the median
volume of nonisotonic intravenous fluid, blood products, or medications administered (Table
S7 in the Supplementary Appendix).

Fewer patients in the balanced-crystalloids group than in the saline group had a measured
Author Manuscript

plasma chloride concentration greater than 110 mmol per liter (24.5% vs. 35.6%, P<0.001)
or a plasma bicarbonate concentration less than 20 mmol per liter (35.2% vs. 42.1%,
P<0.001) (Fig. 2, and Fig. S3 and Table S8 in the Supplementary Appendix). Differences
between groups in chloride and bicarbonate concentration were greater for patients who
received larger volumes of isotonic crystalloid (Figs. S4 and S5 in the Supplementary
Appendix).

PRIMARY OUTCOME
A total of 1139 patients (14.3%) in the balanced-crystalloids group and 1211 patients
(15.4%) in the saline group had a major adverse kidney event (marginal odds ratio, 0.91;
95% confidence interval [CI], 0.84 to 0.99; conditional odds ratio, 0.90; 95% CI, 0.82 to
0.99; P = 0.04) (Table 2, and Table S9 and Fig. S6 in the Supplementary Appendix). The
Author Manuscript

results were similar in six pre-specified sensitivity analyses: one was restricted to patients
who received 500 ml or more of isotonic crystalloid in the 72 hours after enrollment, a
second excluded patients admitted in the week preceding a crossover in the fluid assigned to
the ICU, a third excluded patients who transferred between ICUs or remained in the ICU
through a crossover, a fourth included only the first ICU admission for each patient, a fifth
addressed the issue of missing values for baseline creatinine levels, and a sixth used
alternative modeling approaches (odds ratios between 0.87 and 0.93 for all sensitivity
analyses; see Table S10 in the Supplementary Appendix). In prespecified subgroup analyses,
the difference in the rate of the primary outcome between the balanced-crystalloids group
and the saline group was greater among patients who received larger volumes of isotonic
crystalloid and among patients with sepsis (Fig. 3, and Fig. S7 in the Supplementary
Appendix). Among patients with sepsis, 30-day inhospital mortality was 25.2% with
Author Manuscript

balanced crystalloids and 29.4% with saline (adjusted odds ratio, 0.80; 95% CI, 0.67 to 0.97;
P = 0.02).

SECONDARY OUTCOMES
A total of 818 patients (10.3%) in the balanced-crystalloids group died before hospital
discharge and within 30 days of ICU admission as compared with 875 patients (11.1%) in
the saline group (P = 0.06) (Table 2, and Figs. S8 and S9 in the Supplementary Appendix).

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 7

A total of 189 patients (2.5%) in the balanced-crystalloids group and 220 patients (2.9%) in
Author Manuscript

the saline group received new renal-replacement therapy (P = 0.08) (Table S11 in the
Supplementary Appendix). The highest stage of acute kidney injury and the incidence of
persistent renal dysfunction did not differ significantly between groups (Table 2, and Table
S12 in the Supplementary Appendix).

DISCUSSION
Although both saline and balanced crystalloids have been administered to patients in clinical
practice for decades,3 few trials have addressed the effects of crystalloid composition on
clinical outcomes.1 In preclinical models, the high chloride content of saline has been
reported to cause hyperchloremia,27 acidosis,27 inflammation,28 renal vasoconstriction,29
acute kidney injury,30 hypotension,31 and death.32 Studies involving healthy volunteers
suggest saline may decrease renal perfusion through chloride-mediated renal vaso-
Author Manuscript

constriction.33 Observational studies involving critically ill adults have shown higher rates of
acute kidney injury,34 renal-replacement therapy,5,10 and death6,7,9,35 with saline than with
balanced crystalloids, although results have been inconsistent.36 Although underpowered for
clinical outcomes, two recent pilot trials involving critically ill adults showed an absolute
difference of 1 percentage point in mortality in favor of balanced crystalloids.11,12

In the current trial, the use of balanced crystalloids rather than saline resulted in an absolute
difference of 1.1 percentage points in favor of balanced crystalloids in the primary outcome.
This finding is consistent with the results of the SALT-ED trial conducted concurrently in
noncritically ill adults.14 Although the effect size achieved in the current trial was modest in
terms of percentages, if our data on the use of balanced crystalloids were applied to the care
of the more than 5 million patients admitted to ICUs each year, the reduction in death, new
Author Manuscript

renal-replacement therapy, or persistent renal dysfunction could be substantial.37 Our results

suggest that the use of balanced crystalloids rather than saline might prevent 1 patient among
every 94 patients admitted to an ICU from the need for new renal-replacement therapy, from
persistent renal dysfunction, or from death. Moreover, the difference in outcomes between
balanced crystalloids and saline appeared to be greater for patients with sepsis and patients
who received larger volumes of isotonic crystalloid.

The appropriate composition of a fluid may depend on the indication for its use and the
condition of the individual patient. Concern that the relative hypotonicity of balanced
crystalloids could increase intracranial pressure in patients with brain injury led us to
systematically present clinicians with the option of administering 0.9% sodium chloride to
patients with brain injury, regardless of trial group. Thus, our results cannot be used to
provide guidance as to whether balanced crystalloids should be used in patients with
Author Manuscript

traumatic brain injury.

Our trial has several strengths. The large sample size provided statistical power to detect
small differences in patient outcomes. As was the case in each of the previous trials that
compared balanced crystalloids with saline in critically ill adults,5,11,12 group assignment in
our trial occurred at the level of the ICU. This trial design allowed delivery of the assigned
crystalloid early in each patient’s critical illness. Enrolling all adults admitted to

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 8

participating ICUs and allowing clinical providers to deliver the assigned crystalloid during
Author Manuscript

clinical care minimized selection bias and improved generalizability.

The trial also has several limitations. Conduct at a single academic center limits
generalizability. Treating clinicians were aware of the composition of the assigned
crystalloid and of the group-assignment sequence of their ICU. The outcomes of death and
creatinine level are objective, but a clinician’s decision to initiate renal-replacement therapy
may be susceptible to treatment bias. Censoring data collection at hospital discharge may
underestimate the true incidence of death at 30 days and may overestimate the true incidence
of persistent renal dysfunction at 30 days.16 On the basis of the hypothesized mechanism of
chloride-induced organ injury or acidosis,29,33 we evaluated lactated Ringer’s solution and
Plasma-Lyte A together, and this trial does not inform the choice between the two.

In conclusion, in this trial involving critically ill adults, intravenous administration of

Author Manuscript

balanced crystalloids rather than saline had a favorable effect on the composite outcome of
death, new renal-replacement therapy, or persistent renal dysfunction.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.

Acknowledgments
Supported by the Vanderbilt Institute for Clinical and Translational Research (through grants UL1 TR000445 and
UL1TR002243 from the National Center for Advancing Translational Sciences). Dr. Semler was supported in part
by grants from the National Heart, Lung, and Blood Institute (NHLBI) (HL087738-09 and K12HL133117). Dr. Self
was supported in part by a grant from the National Institute of General Medical Sciences (NIGMS)
(K23GM110469). Dr. Hughes was supported by an American Geriatrics Society Jahnigen Career Development
Award and by grants from the National Institutes of Health (NIH) (HL111111, AG045085, and GM120484). Dr.
Author Manuscript

May was supported in part by grants from the NIGMS (1R01GM115353-01) and the Department of Defense
(12277261). Dr. Casey was supported in part by a grant from the NHLBI (HL087738-09). Dr. Siew was supported
by the Vanderbilt Center for Kidney Disease and the Department of Veterans Affairs’ Health Services Research and
Development Service. Dr. Rice was supported in part by a grant from the NIH (R34HL105869).

We thank the patients, nurses, nurse practitioners, pharmacists, residents, fellows, and attending physicians in the
Vanderbilt Learning Healthcare System for making this trial possible and, in particular, recognize the mentorship of
Arthur P. Wheeler, M.D.

References
1. Myburgh JA, Mythen MG. Resuscitation fluids. N Engl J Med. 2013; 369:1243–51. [PubMed:
24066745]
2. Finfer S, Liu B, Taylor C, et al. Resuscitation fluid use in critically ill adults: an international cross-
sectional study in 391 intensive care units. Crit Care. 2010; 14:R185. [PubMed: 20950434]
Author Manuscript

3. Awad S, Allison SP, Lobo DN. The history of 0. 9% saline. Clin Nutr. 2008; 27:179–88. [PubMed:
18313809]
4. Yunos NM, Kim IB, Bellomo R, et al. The biochemical effects of restricting chloride-rich fluids in
intensive care. Crit Care Med. 2011; 39:2419–24. [PubMed: 21705897]
5. Yunos NM, Bellomo R, Hegarty C, Story D, Ho L, Bailey M. Association between a chloride-liberal
vs chloride-restrictive intravenous fluid administration strategy and kidney injury in critically ill
adults. JAMA. 2012; 308:1566–72. [PubMed: 23073953]

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 9

6. 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; 42:1585–91.
Author Manuscript

[PubMed: 24674927]
7. Rochwerg B, Alhazzani W, Sindi A, et al. Fluid resuscitation in sepsis: a systematic review and
network meta-analysis. Ann Intern Med. 2014; 161:347–55. [PubMed: 25047428]
8. Hammond NE, Taylor C, Finfer S, et al. Patterns of intravenous fluid resuscitation use in adult
intensive care patients between 2007 and 2014: an international cross-sectional study. PLoS One.
2017; 12(5):e0176292. [PubMed: 28498856]
9. Shaw AD, Raghunathan K, Peyerl FW, Munson SH, Paluszkiewicz SM, Schermer CR. Association
between intravenous chloride load during resuscitation and inhospital mortality among patients with
SIRS. Intensive Care Med. 2014; 40:1897–905. [PubMed: 25293535]
10. Shaw AD, Bagshaw SM, Goldstein SL, et al. Major complications, mortality, and resource
utilization after open abdominal surgery: 0. 9% saline compared to Plasma-Lyte. Ann Surg. 2012;
255:821–9. [PubMed: 22470070]
11. 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.
Author Manuscript

JAMA. 2015; 314:1701–10. [PubMed: 26444692]

12. Semler MW, Wanderer JP, Ehrenfeld JM, et al. Balanced crystalloids versus saline in the intensive
care unit: the SALT randomized trial. Am J Respir Crit Care Med. 2017; 195:1362–72. [PubMed:
27749094]
13. Semler MW, Self WH, Wang L, et al. Balanced crystalloids versus saline in the intensive care unit:
study protocol for a cluster-randomized, multiple-crossover trial. Trials. 2017; 18:129. [PubMed:
28302179]
14. Self WH, Semler MW, Wanderer JP, et al. Balanced crystalloids versus saline in noncritically ill
adults. N Engl J Med. 2018; 378:819–28. [PubMed: 29485926]
15. Self WH, Semler MW, Wanderer JP, et al. Saline versus balanced crystalloids for intravenous fluid
therapy in the emergency department: study protocol for a cluster-randomized, multiple-crossover
trial. Trials. 2017; 18:178. [PubMed: 28407811]
16. Semler MW, Rice TW, Shaw AD, et al. Identification of major adverse kidney events within the
electronic health record. J Med Syst. 2016; 40:167. [PubMed: 27234478]
Author Manuscript

17. Shaw A. Models of preventable disease: contrast-induced nephropathy and cardiac surgery-
associated acute kidney injury. Contrib Nephrol. 2011; 174:156–62. [PubMed: 21921620]
18. Palevsky PM, Molitoris BA, Okusa MD, et al. Design of clinical trials in acute kidney injury:
report from an NIDDK workshop on trial methodology. Clin J Am Soc Nephrol. 2012; 7:844–50.
[PubMed: 22442182]
19. Kashani K, Al-Khafaji A, Ardiles T, et al. Discovery and validation of cell cycle arrest biomarkers
in human acute kidney injury. Crit Care. 2013; 17:R25. [PubMed: 23388612]
20. Kellum JA, Zarbock A, Nadim MK. What endpoints should be used for clinical studies in acute
kidney injury? Intensive Care Med. 2017; 43:901–3. [PubMed: 28255614]
21. Závada J, Hoste E, Cartin-Ceba R, et al. A comparison of three methods to estimate baseline
creatinine for RIFLE classification. Nephrol Dial Transplant. 2010; 25:3911–8. [PubMed:
20100732]
22. Kidney Disease Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO
clinical practice guideline for acute kidney injury. Kidney Int. 2012; 2(Suppl):1–138.
23. Parienti J-J, Kuss O. Cluster-crossover design: a method for limiting clusters level effect in
Author Manuscript

community-intervention studies. Contemp Clin Trials. 2007; 28:316–23. [PubMed: 17110172]

24. Turner RM, White IR, Croudace T. Analysis of cluster randomized cross-over trial data: a
comparison of methods. Stat Med. 2007; 26:274–89. [PubMed: 16538700]
25. Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate.
Ann Intern Med. 2009; 150:604–12. [PubMed: 19414839]
26. Shahian DM, Wolf RE, Iezzoni LI, Kirle L, Normand S-LT. Variability in the measurement of
hospital-wide mortality rates. N Engl J Med. 2010; 363:2530–9. [PubMed: 21175315]
27. Kellum JA, Bellomo R, Kramer DJ, Pinsky MR. Etiology of metabolic acidosis during saline
resuscitation in endotoxemia. Shock. 1998; 9:364–8. [PubMed: 9617887]

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 10

28. Kellum JA, Song M, Almasri E. Hyper-chloremic acidosis increases circulating inflammatory
molecules in experimental sepsis. Chest. 2006; 130:962–7. [PubMed: 17035425]
Author Manuscript

29. Wilcox CS. Regulation of renal blood flow by plasma chloride. J Clin Invest. 1983; 71:726–35.
[PubMed: 6826732]
30. Zhou F, Peng Z-Y, Bishop JV, Cove ME, Singbartl K, Kellum JA. Effects of fluid resuscitation with
0. 9% saline versus a balanced electrolyte solution on acute kidney injury in a rat model of sepsis.
Crit Care Med. 2014; 42(4):e270–e278. [PubMed: 24335444]
31. Kellum JA, Song M, Venkataraman R. Effects of hyperchloremic acidosis on arterial pressure and
circulating inflammatory molecules in experimental sepsis. Chest. 2004; 125:243–8. [PubMed:
14718447]
32. Kellum JA. Fluid resuscitation and hyperchloremic acidosis in experimental sepsis: improved
short-term survival and acid-base balance with Hextend compared with saline. Crit Care Med.
2002; 30:300–5. [PubMed: 11889298]
33. 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 148 on renal blood flow
velocity and renal cortical tissue perfusion in healthy volunteers. Ann Surg. 2012; 256:18–24.
Author Manuscript

[PubMed: 22580944]
34. Krajewski ML, Raghunathan K, Paluszkiewicz SM, Schermer CR, Shaw AD. Meta-analysis of
high- versus low-chloride content in perioperative and critical care fluid resuscitation. Br J Surg.
2015; 102:24–36. [PubMed: 25357011]
35. Sen A, Keener CM, Sileanu FE, et al. Chloride content of fluids used for large-volume
resuscitation is associated with reduced survival. Crit Care Med. 2017; 45(2):e146–e153.
[PubMed: 27635770]
36. Rochwerg B, Alhazzani W, Gibson A, et al. Fluid type and the use of renal replacement therapy in
sepsis: a systematic review and network meta-analysis. Intensive Care Med. 2015; 41:1561–71.
[PubMed: 25904181]
37. Wunsch H, Angus DC, Harrison DA, et al. Variation in critical care services across North America
and Western Europe. Crit Care Med. 2008; 36(10):2787–93. e1–9. [PubMed: 18766102]
Author Manuscript
Author Manuscript

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 11
Author Manuscript
Author Manuscript

Figure 1. Volume of Intravenous Isotonic Crystalloid Administered According to Group

The cumulative volume of intravenous balanced crystalloids (solid line) and 0.9% sodium
chloride (dotted line) between admission to the intensive care unit (ICU) and hospital
discharge is shown for patients in the balanced-crystalloids group (Panel A) and the saline
group (Panel B). I bars indicate 95% confidence intervals.
Author Manuscript
Author Manuscript

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 12
Author Manuscript

Figure 2. Plasma Chloride and Bicarbonate Concentration According to Group

Author Manuscript

The mean and 95% confidence interval (denoted by gray shading) for the first measurement
of plasma chloride concentration (Panel A) or bicarbonate concentration (Panel B) on the
first 7 days since admission to the intensive care unit (ICU) are shown for patients in the
balanced-crystalloids group and in the saline group with locally weighted scatterplot
smoothing. Plasma chloride and bicarbonate concentrations were similar between groups at
presentation (Table S3 in the Supplementary Appendix), but because fluid therapy in the
emergency department and operating room was coordinated with the ICU to which patients
were being admitted, plasma chloride concentration differed between the balanced-
crystalloids and saline groups at the time of ICU admission.
Author Manuscript
Author Manuscript

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 13
Author Manuscript
Author Manuscript

Figure 3. Subgroup Analysis of Rates for the Composite Outcome of Death, New Receipt of
Renal-Replacement Therapy, or Persistent Renal Dysfunction
The odds ratio and 95% confidence interval are shown overall and according to subgroup for
the percentage of patients in the balanced-crystalloids group and the saline group who met
the criteria for the composite outcome of death from any cause, new renal-replacement
therapy, or persistent renal dysfunction. Normal kidney function refers to patients who had
no acute kidney injury, chronic kidney disease, or renal-replacement therapy before
Author Manuscript

enrollment. Acute kidney injury refers to patients without chronic kidney disease whose first
creatinine level after enrollment was at least 200% of the baseline value or was both greater
than 4.0 mg per deciliter (350 μmol per liter) and had increased at least 0.3 mg per deciliter
(27 μmol per liter) from the value at baseline.22 Chronic kidney disease refers to patients
with a glomerular filtration rate less than 60 ml per minute per 1.73 m2 as calculated
according to the Chronic Kidney Disease Epidemiology Collaboration equation with the
value for the patient’s baseline creatinine level.25 Previous renal-replacement therapy refers
to patients known to have received any form of renal-replacement therapy before enrollment.
Author Manuscript

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 14

Table 1

Participant Characteristics at Baseline.*

Author Manuscript

Characteristic Balanced Crystalloids (N = 7942) Saline (N = 7860)

Age — yr

Median 58 58

Interquartile range 44–69 44–69

Male sex — no. (%) 4540 (57.2) 4557 (58.0)

White race — no. (%)† 6384 (80.4) 6322 (80.4)

Weight — kg‡

Median 80 79

Interquartile range 69–96 68–95

Coexisting renal conditions — no. (%)

Author Manuscript

Chronic kidney disease of stage 3 or higher§ 1388 (17.5) 1360 (17.3)

Previous receipt of renal-replacement therapy — no. (%) 384 (4.8) 402 (5.1)

Source of admission to ICU — no. (%)

Emergency department 3975 (50.1) 3997 (50.9)

Operating room 1732 (21.8) 1649 (21.0)

Transfer from another hospital 1038 (13.1) 1018 (13.0)

Hospital ward 788 (9.9) 780 (9.9)

Outpatient 363 (4.6) 359 (4.6)

Another ICU within hospital 46 (0.6) 57 (0.7)

Diagnosis on ICU admission — no. (%)

Sepsis or septic shock 1167 (14.7) 1169 (14.9)

Author Manuscript

Traumatic brain injury 698 (8.8) 665 (8.5)

Mechanical ventilation — no. (%) 2723 (34.3) 2731 (34.7)

Vasopressors — no. (%) 2094 (26.4) 2058 (26.2)

Mean predicted risk of in-hospital death — % (95% CI)¶ 9.4 (9.0–9.9) 9.6 (9.2–10.0)

Baseline creatinine level — mg/dl||

Median 0.89 0.89

Interquartile range 0.74–1.10 0.74–1.10

Acute kidney injury of stage 2 or higher — no. (%)** 681 (8.6) 643 (8.2)

*
There were no significant differences in baseline characteristics between the two study groups (P values range from 0.12 to 0.94). To convert the
values for creatinine to micromoles per liter, multiply by 88.4. ICU denotes intensive care unit.
†
Race was reported by patients or their surrogates and recorded in the electronic health record as a part of routine clinical care.
Author Manuscript

‡
Information on weight at enrollment was missing for 698 patients.
§
Chronic kidney disease of stage 3 or higher is defined as a glomerular filtration rate less than 60 ml per minute per 1.73 m2, as calculated with the
equation developed by the Chronic Kidney Disease Epidemiology Collaboration25 with the patient’s baseline creatinine value.

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 15

¶
Predicted risk of in-hospital death is an estimated probability of death before hospital discharge generated through the Vizient database (formerly
known as the University HealthSystem Consortium).26 Information on the predicted risk of in-hospital death was missing for 126 patients.
||
For the purposes of the trial, the baseline creatinine level was defined as the lowest plasma creatinine level measured in the 12 months preceding
Author Manuscript

hospitalization, unless not available, in which case the lowest plasma creatinine level measured between hospitalization and admission to the ICU
was used. An estimated creatinine level was used for patients for whom there was no level available from the 12 months before hospitalization to
the time of admission to the ICU. Baseline creatinine levels were estimated for a total of 863 patients (10.9%) in the balanced-crystalloids group
and 826 patients (10.5%) in the saline group (Table S3 in the Supplementary Appendix).
**
Acute kidney injury of stage 2 or higher is defined according to the Kidney Disease: Improving Global Outcomes creatinine criteria22 as a first
plasma creatinine value after enrollment of at least 200% of the baseline value or both a value greater than 4.0 mg per deciliter (350 μmol per liter)
and an increase of at least 0.3 mg per deciliter (27 μmol per liter) from the baseline value.
Author Manuscript
Author Manuscript
Author Manuscript

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 16

Table 2

Clinical Outcomes.*
Author Manuscript

Balanced Crystalloids Adjusted Odds

Outcome (N = 7942) Saline (N = 7860) Ratio (95% CI)† P Value†
Primary outcome

Major adverse kidney event within 30 days — no. (%)‡ 1139 (14.3) 1211 (15.4) 0.90 (0.82 to 0.99) 0.04

Components of primary outcome

In-hospital death before 30 days — no. (%) 818 (10.3) 875 (11.1) 0.90 (0.80 to 1.01) 0.06

Receipt of new renal-replacement therapy — no./total no. 189/7558 (2.5) 220/7458 (2.9) 0.84 (0.68 to 1.02) 0.08
(%)§

Among survivors 106/6787 (1.6) 117/6657 (1.8)

Final creatinine level ≥200% of baseline — no./total no. 487/7558 (6.4) 494/7458 (6.6) 0.96 (0.84 to 1.11) 0.60
(%)§
Author Manuscript

Among survivors 259/6787 (3.8) 273/6657 (4.1)

Among survivors without new renal-replacement therapy 215/6681 (3.2) 219/6540 (3.3)

Secondary outcomes
In-hospital death — no. (%)

Before ICU discharge 528 (6.6) 572 (7.3) 0.89 (0.78 to 1.02) 0.08

Before 60 days 928 (11.7) 975 (12.4) 0.92 (0.83 to 1.02) 0.13

ICU-free days¶ 0.94

Median 25.3 25.3 1.00 (0.89 to 1.13)

Interquartile range 22.1 to 26.6 22.2 to 26.6

Mean 21.8±8.3 21.7±8.6

Ventilator-free days¶ 1.06 (0.97 to 1.16) 0.22

Author Manuscript

Median 28.0 28.0

Interquartile range 26.0 to 28.0 26.0 to 28.0

Mean 24.2±8.6 23.9±8.9

Vasopressor-free days¶ 1.05 (0.97 to 1.14) 0.26

Median 28.0 28.0

Interquartile range 27.0 to 28.0 27.0 to 28.0

Mean 24.7±8.5 24.4±8.8

Renal-replacement therapy–free days¶ 1.11 (1.02 to 1.20) 0.01

Median 28.0 28.0

Interquartile range 28.0 to 28.0 28.0 to 28.0

Mean 25.0±8.6 24.8±8.9

Author Manuscript

Secondary renal outcomes§

Stage 2 or higher AKI developing after enrollment — no./ 807/7558 (10.7) 858/7458 (11.5) 0.91 (0.82 to 1.01) 0.09
total no. (%)||

Creatinine — mg/dl**

Highest before discharge or day 30 1.01 (0.97 to 1.05) 0.58

N Engl J Med. Author manuscript; available in PMC 2018 September 01.

 Semler et al. Page 17

Balanced Crystalloids Adjusted Odds

Outcome (N = 7942) Saline (N = 7860) Ratio (95% CI)† P Value†
Author Manuscript

Median 0.99 0.99

Interquartile range 0.78 to 1.53 0.78 to 1.52

Change from baseline to highest value 0.98 (0.94 to 1.02) 0.35

Median 0.04 0.04

Interquartile range −0.08 to 0.31 −0.08 to 0.32

Final value before discharge or 30 days 1.02 (0.97 to 1.06) 0.51

Median 0.83 0.83

Interquartile range 0.70 to 1.11 0.70 to 1.11

*
Plus–minus values are means ±SD. To convert the values for creatinine to micromoles per liter, multiply by 88.4. ICU denotes intensive care unit.
†
Categorical outcomes were compared with a generalized, linear, mixed-effects model, with adjustment for the ICU to which the patient was
admitted as a random effect and prespecified covariates as fixed effects.13 Continuous outcomes were compared between groups with a
Author Manuscript

proportional-odds model, with adjustment for the same variables.

‡
A major adverse kidney event within 30 days is the composite of death, receipt of new renal-replacement therapy, or final creatinine level that was
at least 200% of the baseline level, with all events censored at hospital discharge or at 30 days after admission to the ICU, whichever occurred first.
The effect of study group on major adverse kidney events within 30 days is the conditional effect. The marginal effect yielded an odds ratio of 0.91
and a 95% confidence interval of 0.84 to 0.99.
§
Data on receipt of new renal-replacement therapy, final creatinine level that was at least 200% of the baseline level, and secondary renal outcomes
are provided for the 15,016 patients not known to have received renal-replacement therapy before ICU admission.
¶
ICU-free, ventilator-free, vasopressor-free, and renal-replacement-therapy–free days refer to the number of days on which a patient was alive and
free from the specified therapy in the first 28 days after enrollment. Odds ratios of higher than 1.0 indicate a better outcome (i.e., more days alive
and free from the specified therapy) with balanced crystalloids than with saline.
||
The development of acute kidney injury (AKI) of stage 2 or higher after enrollment was defined in accordance with the Kidney Disease:
Improving Global Outcomes plasma creatinine criteria22 as any creatinine level between enrollment and discharge or 30 days that increased by at
least 0.3 mg per deciliter (27 μmol per liter) from a preceding post-enrollment value and was at least 200% of the baseline value, at least 200% of a
Author Manuscript

preceding post-enrollment value, or at least 4.0 mg per deciliter (350 μmol per liter) or as new receipt of renal-replacement therapy.
**
Among patients who had not received previous renal-replacement therapy, the plasma creatinine level was measured a mean of 8.0 times between
enrollment and the first of discharge or 30 days in each group; the plasma creatinine level was not measured between enrollment and the first of
discharge or 30 days for 418 of 7558 patients (5.5%) in the balanced-crystalloids group and 443 of 7458 patients (5.9%) in the saline group.
Author Manuscript

N Engl J Med. Author manuscript; available in PMC 2018 September 01.