Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis J Peds 2017 PDF
Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis J Peds 2017 PDF
Objective To test the hypothesis that resuscitation with balanced fluids (lactated Ringer [LR]) is associated with
improved outcomes compared with normal saline (NS) in pediatric sepsis.
Study design We performed matched analyses using data from 12 529 patients <18 years of age with severe
sepsis/septic shock at 382 US hospitals between 2000 and 2013 to compare outcomes with vs without LR as part
of initial resuscitation. Patients receiving LR were matched 1:1 to patients receiving only NS (NS group), including
separate matches for any (LR-any group) or exclusive (LR-only group) LR use. Outcomes included 30-day hos-
pital mortality, acute kidney injury, new dialysis, and length of stay.
Results The LR-any group was older, received larger crystalloid volumes, and was less likely to have malignan-
cies than the NS group. After matching, mortality was not different between LR-any (7.2%) and NS (7.9%) groups
(risk ratio 0.99, 95% CI 0.98, 1.01; P = .20). There were no differences in secondary outcomes except longer hos-
pital length of stay in LR-any group (absolute difference 2.4, 95% CI 1.4, 5.0 days; P < .001). Although LR was
preferentially used as adjunctive fluid with large-volume resuscitation or first-line fluid in patients with lower illness
severity, outcomes were not different after matching stratified by volume and proportionate LR utilization, including
for patients in the LR-only group.
Conclusions Balanced fluid resuscitation with LR was not associated with improved outcomes compared with NS
in pediatric sepsis. Although the current practice of NS resuscitation is justified, selective LR use necessitates a pro-
spective trial to definitively determine comparative effectiveness among crystalloids. (J Pediatr 2016;■■:■■-■■).
F
luid resuscitation is the cornerstone of acute management for hypovolemia and shock, but there remains uncertainty as
to the most appropriate fluid to restore blood volume and optimize organ perfusion.1-3 Isotonic crystalloid fluids are
generally preferred, except in cases of hemorrhage, as they are inexpensive, easy to store, and available in a wide variety
of settings.4,5 Sepsis guidelines for adults and pediatrics recommend initial crystalloid fluid resuscitation.6,7
Crystalloid fluids can be categorized as either nonbuffered/nonbalanced (eg, 0.9% normal saline [NS]) or balanced (eg, lac-
tated Ringer [LR], Hartmann, Plasma-Lyte, Baxter, Deerfield, Illinois) solutions. Although balanced fluid have a more physi-
ologic electrolyte composition and strong ion difference closer to plasma than NS, these fluids have not been preferentially used
for sepsis resuscitation.4,5,8 However, large amounts of NS can induce a hyperchloremic metabolic acidosis and have been as-
sociated with adverse effects on kidney injury, coagulation, and death.9-12 Alternatively, balanced crystalloids have been associ-
ated with improved outcomes and decreased renal replacement therapy compared with NS in adult sepsis.11,13
In pediatric sepsis, there are limited data comparing clinical outcomes follow-
ing LR vs NS resuscitation. Although Carcillo et al14 demonstrated the impor-
tance of early fluid resuscitation in pediatric septic shock, there was no differentiation
between use of NS or LR. In a randomized trial of 4 fluid regimens in children
with dengue fever, patients receiving LR were slower to recover from shock com- From the 1Division of Critical Care Medicine, Department
of Anesthesiology and Critical Care, The Children’s
pared with NS, but the study was not powered for morbidity or mortality Hospital of Philadelphia, University of Pennsylvania
Perelman School of Medicine, Philadelphia, PA; 2McCourt
outcomes.15 The largest study of fluid resuscitation in children with severe infec- School of Public Policy and Department of Government,
Georgetown University, Washington, DC; 3Center for
tions restricted crystalloid fluids to NS.16 Consequently, guidelines for pediatric Pediatric Clinical Effectiveness, The Children’s Hospital of
sepsis are unable to provide evidence-based recommendations to choose among Philadelphia, Philadelphia, PA; 4Division of Emergency
Medicine; and 5Division of Infectious Diseases,
available crystalloid solutions even despite emerging data questioning the rela- Department of Pediatrics, The Children’s Hospital of
Philadelphia, University of Pennsylvania Perelman School
tive safety of NS in adults.6 Because crystalloid fluids are so commonly used, even of Medicine, Philadelphia, PA
Supported by the Department of Anesthesiology and
Critical Care, Division of Emergency Medicine, and
Center for Pediatric Clinical Effectiveness at The
Children’s Hospital of Philadelphia. S.W. receives support
from National Institute of General Medical Sciences
AKI Acute kidney injury (K23GM110496). F.B. receives support from the Eunice
ICD-9-CM International Classification of Diseases, Ninth Edition, Clinical Modification Kennedy Shriver National Institute of Child Health and
LOS Length of stay Human Development (K23HD082368). The authors
declare no conflicts of interest.
LR Lactated Ringer
NS Normal saline 0022-3476/$ - see front matter. © 2016 Elsevier Inc. All rights
PICU Pediatric intensive care unit reserved.
http://dx.doi.org10.1016/j.jpeds.2016.11.075
a small benefit attributable to type of fluid resuscitation could Demographics, month/site of admission, comorbid condi-
provide a substantial public health impact with only a minor tions, and intensive care therapies were obtained from the
shift in practice. We, therefore, sought to test the hypothesis Premier Healthcare Database. Comorbid conditions were
that balanced fluid resuscitation is associated with improved defined using pediatric complex chronic conditions.19 Thera-
outcomes in pediatric sepsis. pies included use of the following on hospitals days 1, 2, or
3: noninvasive and invasive mechanical ventilation, vasoac-
tive infusions, albumin, blood products, furosemide, cortico-
Methods steroids, use of a central venous catheter, arterial line, or bladder
catheter, and extracorporeal membrane oxygenation. Because
We conducted a matched retrospective cohort study of pedi- doses of vasoactive infusions were not available, we summa-
atric patients <18 years of age with severe sepsis or septic shock rized this variable as the total number of vasoactive infu-
across 382 geographically diverse US hospitals between January sions. Blood products were defined as any combination of red
2000 and December 2013. Patients were identified from the blood cells, platelets, fresh frozen plasma, or cryoprecipitate.
Premier Healthcare Database, an administrative database es-
tablished by the Premier healthcare alliance that contains item- Outcomes
ized daily logs of all patient charges. The Premier Healthcare The primary outcome was all-cause 30-day hospital mortal-
Database is the largest acute care database in the US with a ity in the NS vs LR-any groups. To increase the likelihood that
complete census of all inpatients from more than 600 hospi- death was related to the initial sepsis episode requiring fluid
tals, of which approximately three-quarters are nonteaching resuscitation, we censored the primary outcome at 30 days after
hospitals. Pediatric data is contributed through a combina- admission. Secondary outcomes included uncensored hospi-
tion of community-based and specialty children’s hospitals. tal mortality, hospital mortality plus hospice, acute kidney injury
The study was considered exempt from human subjects re- (AKI) with and without dialysis, and pediatric intensive care
search oversight by The Children’s Hospital of Philadelphia unit (PICU) and hospital length of stay (LOS). AKI was defined
Institutional Review Board because only deidentified data were by the ICD-9-CM code 584.x and AKI with dialysis was defined
used. as an ICD-9-CM code for AKI (584.x) with either (a) a pro-
Eligible patients were <18 years of age, diagnosed with severe cedure charge for a dialysis catheter (38.95) with a charge for
sepsis or septic shock, received initial treatment at the Premier dialysis (39.95) or (b) charge codes for dialysis supplies.13 Pa-
hospital, were not admitted to a neonatal intensive care unit tients with an ICD-9-CM code for end-stage renal disease
(based on all patient refined-diagnosis related group codes), already undergoing dialysis (ICD-9-CM 585.6) were ex-
and were ordered to receive any combination of NS or LR fluid cluded from the analysis of AKI with or without dialysis. All
boluses during the first 3 days of hospital admission. To iden- outcomes were also analyzed separately for patients in the LR-
tify severe sepsis and septic shock, we used previously pub- only group.
lished combinations of International Classification of Diseases,
Ninth Edition, Clinical Modification (ICD-9-CM) codes for Statistical Analyses
either an invasive infection plus acute organ dysfunction Analyses were performed using R 2.13.1 (R Foundation) with
(Tables I and II; available at www.jpeds.com) or the ICD-9- mipmatch package20 and Stata v 12.1 (StataCorp, College
CM codes for severe sepsis (785.52) or septic shock (995.92).17,18 Station, Texas). Data are presented as medians (IQR) or pro-
To increase the likelihood that initial fluid resuscitation was portions. We used mixed integer programming 1:1 matching
related to sepsis, we restricted inclusion to patients with blood to minimize the within-pair Mahalanobis distance for key
cultures and broad-spectrum antibiotics (Table III; available covariates that were both available within Premier and had a
at www.jpeds.com) ordered within the first 3 hospital days. We biologically plausible or previously demonstrated associa-
excluded patients with unknown hospital disposition at day tion, including demographics, comorbidities, and therapies, with
30. risk of death. The Mahalanobis distance is the difference in
Exposure to LR or NS was defined by type and amount of covariate values for patients in the LR vs NS groups divided
fluid recorded over the first three hospital days. Only LR or by the covariates’ SD.21 Unlike propensity scores that can
NS ordered as bolus therapy was considered. Because bal- produce stochastic balance, integer matching ensures a more
anced fluids other than LR (eg, Plasma-Lyte) were rare predictable and precise balance on specific covariates.20 The
(0.3%), we limited our analysis to LR and NS. Fluid volumes specific patient-level covariates used for matching are listed in
were billed as 250, 500, or 1000 mL units. Although some Tables IV and V (Table IV; available at www.jpeds.com). In
patients likely received only a portion of a unit because of addition, because LR use was likely to cluster by hospital, we
weight-based fluid dosing in pediatrics, we considered the also matched within site exactly except for hospitals that had
entire unit to have been administered. Patients were catego- ≤10 patients for which we allowed matching across sites. We
rized as exposure to only NS (NS group) or to varying also repeated the analysis excluding hospitals with ≤10 pa-
amounts LR and NS (LR-any group), similar to the method- tients to ensure matching across low-volume hospitals did not
ology published by Raghunathan et al.13 We also performed impact our findings. Because prior studies have demon-
a separate analysis of patients who received only NS vs only strated differences in mortality for patients identified with spe-
LR (LR-only group). cific severe sepsis/septic shock ICD-9-CM codes compared with
2 Weiss et al
Table VII. Outcomes in matched cohorts for LR-any and LR-only groups
LR-any group† NS group‡ Risk ratio/difference
Outcomes* (n = 2117) (n = 2117) (95% CI) P value§
Mortality, 30-d 153 (7.2) 168 (7.9) 0.99 (0.9, 1.09) .20
Mortality, hospital 194 (9.2) 199 (9.4) 1.0 (0.98, 1.02) .41
Mortality, including hospice 200 (9.4) 211 (10.0) 0.99 (0.98, 1.01) .29
AKI 334 (15.8) 337 (15.9) 1.0 (0.97, 1.02) .41
New dialysis 27 (1.3) 33 (1.6) 1.0 (0.99, 1.00) .21
PICU LOS¶, median (IQR) 7.8 (1.0, 13.0) 7.3 (1.0, 12.0) 0.5 (0.2, 0.8) .01
Hospital LOS¶, median (IQR) 15.5 (6.0, 22.0) 13.1 (4.0, 20.0) 2.4 (1.4, 5.0) <.001
tients included in the matched analysis were identified with Volume-Stratified Analysis
severe sepsis using ICD-9-CM codes for infection plus organ Although mortality rose with increasing weight-adjusted crys-
dysfunction (80.4%) vs specific severe sepsis/septic shock codes talloid fluid volume, 30-day hospital mortality did not differ
(19.6%). between the LR-any and NS groups after matching within
volume quartiles (Figure 4). However, only 142 patients re-
LR-Any Matched Analysis ceived LR in the first quartile of total crystalloid fluid volume,
In the matched cohort of 4234 patients, 30-day hospital mor- and there was an increase in LR utilization with successive
tality was 7.2% in the LR-any group and 7.9% in the NS group
(risk ratio 0.99, 95% CI 0.98, 1.01; P = .20). There were no sig-
nificant differences in overall hospital mortality, hospital mor-
tality plus hospice, or AKI with and without dialysis (Table VII).
There remained no differences in these outcomes after further
adjusting for imperfectly balanced covariates in multivari-
able analyses, after repeating the match using only those
covariates unlikely to mediate the effect of fluid type on
outcome, or after excluding surgical patients (data not shown).
Hospital LOS was longer for the LR-any group compared with
the NS group (absolute difference 2.4, 95% CI 1.4, 5.0 days;
P < .001). Analyses setting LOS to the sample maximum or to
30 days for all nonsurvivors did not impact these findings
(Table VIII; available at www.jpeds.com), nor did excluding
234 matched pairs from hospitals with ≤10 patients (data not
shown). The Kaplan-Meier analysis demonstrated no differ- Figure 4. Hospital mortality for LR-any and NS groups matched
ence in time-to-death between matched groups (log-rank within quartile of total crystalloid fluid volume. The x-axis cat-
P = .11) (Figure 3; available at www.jpeds.com). egorizes patients based on quartile of total fluid crystalloid
Patients with specific ICD-9-CM codes for severe sepsis/ volume received after correcting by estimated weight for age
septic shock had higher 30-day hospital mortality than pa- (median total volume in Q1 = 8 mL/kg, Q2 = 17 mL/kg,
tients identified by infection plus organ dysfunction codes Q3 = 32 mL/kg, and Q4 = 68 mL/kg). The y-axis shows the ad-
(15.0% vs 5.8%, P < .001). However, there were no differ- justed 30-day mortality rate. Patients in the LR-any group were
ences in outcomes between the matched LR-any and NS groups matched within volume quartile to patients who received only
stratified by sepsis identification strategy except for longer PICU NS. There were no significant differences in mortality between
the LR-any and NS groups within any of the total volume
and hospital LOS for the LR group identified by combina-
quartiles.
tion codes (Table IX; available at www.jpeds.com).
4 Weiss et al
quartiles of total volume administration (Q1: 5.0%; Q2: 11.5%; rich solutions can also cause a hyperchloremic metabolic
Q3: 20.6%; Q4: 30.2%), supporting the preferential use of LR acidosis, which has been shown to be proinflammatory.30
in patients who required larger total fluid volumes. In addi- Our work builds on previous clinical studies that have over-
tion, PICU admission and the use of nonfluid intensive thera- whelmingly focused on adult populations. Most studies have
pies were more common in each successive quartile (Table X; demonstrated either no difference31-34 or a benefit of at least
available at www.jpeds.com) making it difficult to disen- some proportion of resuscitation fluids given as balanced
tangle proportionate LR use, volume of fluid resuscitation, and solutions.11-13 For example, Raghunathan et a13 found that receipt
illness severity. Finally, matching performed least effectively in of at least some balanced fluids during initial resuscitation was
the fourth quartile of patients who received the largest total associated with lower hospital mortality in adults with
fluid volumes and with the most severe illness severity (Figure 5; vasopressor-dependent septic shock. This contrast with our
available at www.jpeds.com). results may be attributable to age-related biological differ-
ences, such as a lower rate of baseline subclinical cardiac and
Dose-Response Analysis renal disease in children, or methodological differences, such
To account for variability in the proportionate use of LR, pa- as limiting the adult study to vasopressor-dependent shock.
tients were separately matched after stratifying by propor- Unfortunately, insufficient sample size precluded limiting our
tion of total crystalloid volume ordered as LR. Patients with pediatric analysis to vasopressor-dependent shock. Moreover,
an increasing proportionate LR use received fewer intensive even though Premier offers a geographically diverse sample,
therapies and had a lower rate of adverse outcomes (Tables XI the relatively low median fluid volume resuscitation and mor-
and XII; available at www.jpeds.com). However, there were no tality <8% suggests an overall moderate illness severity that may
differences in 30-day hospital mortality or AKI (Figure 6; avail- not reflect more severe pediatric sepsis cases that tend to con-
able at www.jpeds.com) or dialysis (data not shown) between centrate at specialty children’s hospitals. It is also possible that
the LR groups stratified by proportionate LR utilization and potential detrimental effects of LR, including microvascular
the NS group. thromboses (because of calcium activating the clotting cascade)9
or cerebral edema (because of mild fluid hypotonicity),35 may
LR-Only Matched Analysis be more problematic in children. For example, in Vietnamese
In the separate matched cohort of 918 patients receiving either children with Dengue shock, patients randomized to resusci-
exclusive LR or NS fluid, 30-day hospital mortality was 5% in tation with LR had a longer time to recovery than patients re-
the LR-only group and 4.6% in the NS group (risk ratio 1.01, suscitated with NS.15 As with other critical therapies, benefits
95% CI 0.98, 1.03; P = .69). There were no significant differ- seen in adult populations may not translate to children.
ences in secondary outcomes, except a longer hospital LOS in Prior studies suggest that potential adverse effects associ-
the LR-only group (Table VII). ated with NS are dose-related such that LR may only be ben-
eficial for patients requiring large-volume fluid resuscitation.1-3
Discussion In our study, mortality increased with larger fluid volumes and
decreased with a greater proportion of fluid given as LR, but
In this large matched cohort study of pediatric severe sepsis there were no differences between LR and NS groups after
and septic shock, balanced fluid resuscitation with LR was not matching within volume quartiles, by proportionate LR uti-
associated with improved mortality, AKI, or dialysis, even when lization, or in the separate matched analysis of LR-only pa-
matched by fluid volume and proportionate LR utilization. tients. However, the preference for LR as first-line fluid in
However, LR was preferentially used either as first-line fluid patients with low illness severity or as an adjunctive fluid in
in patients with lower illness severity or as an adjunctive fluid patients who received a large amount of total fluid could have
in patients who received large amounts of fluid resuscita- masked a true benefit of LR.
tion, and the matching algorithm was least effective in the most There are several limitations. First, claims-based data may
severely ill patients who received the largest total fluid volumes. lead to misclassification bias if ordered and administered thera-
Consequently, the results of our study are best interpreted as pies are discrepant, although this is unlikely to produce dif-
establishing the need for and the equipoise to conduct a pro- ferential bias between groups. Also, we were not able to account
spective randomized trial to definitely address the compara- for prehospital fluid administration, partial administration of
tive effectiveness of balanced fluids and NS in pediatric sepsis. fluid units (which is common in pediatrics because of weight-
Prior data suggest that the supraphysiologic chloride content based fluid administration), or account for deviations in weight
of NS may be detrimental to renal function and acid-base from published growth curves. Even though this may have over-
balance.1-3,9-11 Infusion of chloride-rich fluids reduced renal or underestimated fluid administration, these are unlikely to
blood flow in dogs and healthy human volunteers to a greater have been sources of differential bias between LR and NS.
extent than more balanced fluids.27,28 NS also induced ab- However, the younger age of the NS-only group (before and
dominal discomfort, drowsiness, and impaired cognition, com- after matching) could have introduced slightly more error in
pared with LR and other balanced fluids, in a human study.29 estimated total fluid volumes than in the LR-group because
In a sequential period study of critically ill adults, use of younger patients are more likely to receive partial adminis-
chloride-restrictive fluids reduced the odds of AKI and dialy- tration of a fluid bag and rounding to median weight repre-
sis by almost 50%.11 Infusion of large volumes of chloride- sents a larger proportional change relative to true weight.
Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis: A Matched Retrospective Cohort Study 5
Second, using ICD-9-CM codes for infection plus organ dys- 6. Brierley J, Carcillo JA, Choong K, Cornell T, Decaen A, Deymann A, et al.
function to identify pediatric sepsis is controversial.17,18,36 Clinical practice parameters for hemodynamic support of pediatric and
neonatal septic shock: 2007 update from the American College of Criti-
Notably, patients with more specific ICD-9-CM sepsis codes cal Care Medicine. Crit Care Med 2009;37:666-88.
had a trend toward decreased mortality and less dialysis in the 7. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, et al.
LR group. Moreover, because identification and timing of sur- Surviving sepsis campaign: international guidelines for management of
gical interventions is limited using administrative codes, it was severe sepsis and septic shock: 2012. Crit Care Med 2013;41:580-
difficult to fully account for a possible surgical preference to 637.
8. Ventura AM, Shieh HH, Bousso A, Goes PF, de Cassia FOFI, de Souza
use LR. Third, differences in demographics, comorbidities, and DC, et al. Double-blind prospective randomized controlled trial of do-
intensive therapies indicate nonrandom selective use of LR over pamine versus epinephrine as first-line vasoactive drugs in pediatric septic
NS. Although statistical matching was able to remove much shock. Crit Care Med 2015;43:2292-302.
of these baseline differences, we cannot rule out residual con- 9. Kiraly LN, Differding JA, Enomoto TM, Sawai RS, Muller PJ, Diggs B,
founding within unmeasured covariates. In addition, in the et al. Resuscitation with normal saline (NS) vs. lactated ringers (LR) modu-
lates hypercoagulability and leads to increased blood loss in an uncon-
absence of physiologic and laboratory data, we used inten- trolled hemorrhagic shock swine model. J Trauma 2006;61:57-64, discussion
sive care therapies for illness severity but could not deter- -5.
mine if these therapies occurred after—or as a result 10. Zhou F, Peng ZY, Bishop JV, Cove ME, Singbartl K, Kellum JA. Effects of
of—differential use of LR vs NS. Because excluding these vari- fluid resuscitation with 0.9% saline versus a balanced electrolyte solution
ables from the match posed risk for increased confounding, on acute kidney injury in a rat model of sepsis. Crit Care Med 2014;42:e270-
8.
we chose a conservative approach similar to Raghunathan et al13 11. Yunos NM, Bellomo R, Hegarty C, Story D, Ho L, Bailey M. Association
fluid study in adult sepsis despite the possibility that match- between a chloride-liberal vs chloride-restrictive intravenous fluid ad-
ing on these covariates could have masked outcome differ- ministration strategy and kidney injury in critically ill adults. JAMA
ences. However, secondary analyses using patients matched only 2012;308:1566-72.
on covariates unlikely to mediate the effect of fluid type on 12. Shaw AD, Bagshaw SM, Goldstein SL, Scherer LA, Duan M, Schermer CR,
et al. Major complications, mortality, and resource utilization after open
outcome also showed no differential effect of LR vs NS. Finally, abdominal surgery: 0.9% saline compared to Plasma-Lyte. Ann Surg
because LR was the predominant balanced fluid used, our data 2012;255:821-9.
may not be generalizable to other balanced fluids. 13. Raghunathan K, Shaw A, Nathanson B, Sturmer T, Brookhart A, Stefan
In this large matched observational study, use of LR (alone MS, et al. Association between the choice of IV crystalloid and in-
or in combination with NS) was not associated with im- hospital mortality among critically ill adults with sepsis. Crit Care Med
2014;42:1585-91.
proved outcomes compared with exclusive NS resuscitation in 14. Carcillo JA, Davis AL, Zaritsky A. Role of early fluid resuscitation in pe-
pediatric septic shock. These findings support the current prac- diatric septic shock. JAMA 1991;266:1242-5.
tice of using NS as the first choice for crystalloid fluid resus- 15. Ngo NT, Cao XT, Kneen R, Wills B, Nguyen VM, Nguyen TQ, et al. Acute
citation in pediatric sepsis. However, given the limitations of management of dengue shock syndrome: a randomized double-blind com-
matching within a retrospective observational study to fully parison of 4 intravenous fluid regimens in the first hour. Clin Infect Dis
2001;32:204-13.
account for the nonrandom selective use of LR, our findings 16. Maitland K, Kiguli S, Opoka RO, Engoru C, Olupot-Olupot P, Akech SO,
also emphasize the need for a large-scale prospective random- et al. Mortality after fluid bolus in African children with severe infection.
ized trial to definitely determine the comparative effective- N Engl J Med 2011;364:2483-95.
ness of balanced fluids and NS in pediatric sepsis. ■ 17. Balamuth F, Weiss SL, Neuman MI, Scott H, Brady PW, Paul R, et al. Pe-
diatric severe sepsis in U.S. children’s hospitals. Pediatr Crit Care Med
2014;15:798-805.
Submitted for publication Jul 21, 2016; last revision received Sep 28, 2016;
18. Weiss SL, Parker B, Bullock ME, Swartz S, Price C, Wainwright MS, et al.
accepted Nov 29, 2016
Defining pediatric sepsis by different criteria: discrepancies in popula-
Reprint requests: Scott L. Weiss, MD, MSCE, Department of Anesthesiology tions and implications for clinical practice. Pediatr Crit Care Med
and Critical Care, The Children’s Hospital of Philadelphia, University of
2012;13:e219-26.
Pennsylvania Perelman School of Medicine, 3401 Civic Center Blvd, 7 South
Tower, Room 7C04, Philadelphia, PA 19104. E-mail: [email protected] 19. Feudtner C, Hays RM, Haynes G, Geyer JR, Neff JM, Koepsell TD. Deaths
attributed to pediatric complex chronic conditions: national trends
and implications for supportive care services. Pediatrics 2001;107:
References E99.
20. Zubizarreta JR. Using mixed integer programming for matching in an ob-
1. Karakala N, Raghunathan K, Shaw AD. Intravenous fluids in sepsis: what servational study of kidney failure after surgery. J Am Stat Assoc
to use and what to avoid. Curr Opin Crit Care 2013;19:537-43. 2012;107:1360-71.
2. Santi M, Lava SA, Camozzi P, Giannini O, Milani GP, Simonetti GD, et al. 21. Rubin DB. Bias reduction using Mahalanobis-metric matching. Biomet-
The great fluid debate: saline or so-called “balanced” salt solutions? Ital rics 1980;36:293-8.
J Pediatr 2015;41:47. 22. Rosenbaum PR. Sensitivity analysis for m-estimates, tests, and confi-
3. Myburgh JA, Mythen MG. Resuscitation fluids. N Engl J Med dence intervals in matched observational studies. Biometrics 2007;63:456-
2013;369:1243-51. 64.
4. Boulain T, Boisrame-Helms J, Ehrmann S, Lascarrou JB, Bougle A, Chiche 23. Neuman MD, Rosenbaum PR, Ludwig JM, Zubizarreta JR, Silber JH. An-
A, et al. Volume expansion in the first 4 days of shock: a prospective esthesia technique, mortality, and length of stay after hip fracture surgery.
multicentre study in 19 French intensive care units. Intensive Care Med JAMA 2014;311:2508-17.
2015;41:248-56. 24. Silber JH, Rosenbaum PR, Trudeau ME, Even-Shoshan O, Chen W, Zhang
5. Cecconi M, Hofer C, Teboul JL, Pettila V, Wilkman E, Molnar Z, et al. X, et al. Multivariate matching and bias reduction in the surgical out-
Fluid challenges in intensive care: the FENICE study: a global inception comes study. Med Care 2001;39:1048-64.
cohort study. Intensive Care Med 2015;41:1529-37. 25. Pearce N. Analysis of matched case-control studies. BMJ 2016;352:i969.
6 Weiss et al
26. Lin W, Halpern SD, Prasad Kerlin M, Small DS. A “placement of death” critically ill patients presenting with hypovolemic shock: the CRISTAL
approach for studies of treatment effects on ICU length of stay. Stat Methods randomized trial. JAMA 2013;310:1809-17.
Med Res 2014 Aug 1; pii: 0962280214545121. [Epub ahead of print]. 32. Rochwerg B, Alhazzani W, Gibson A, Ribic CM, Sindi A, Heels-Ansdell
27. Chowdhury AH, Cox EF, Francis ST, Lobo DN. A randomized, con- D, et al. Fluid type and the use of renal replacement therapy in sepsis: a
trolled, double-blind crossover study on the effects of 2-L infusions of systematic review and network meta-analysis. Intensive Care Med
0.9% saline and plasma-lyte(R) 148 on renal blood flow velocity and renal 2015;41:1561-71.
cortical tissue perfusion in healthy volunteers. Ann Surg 2012;256:18- 33. Rochwerg B, Alhazzani W, Sindi A, Heels-Ansdell D, Thabane L, Fox-
24. Robichaud A, et al. Fluid resuscitation in sepsis: a systematic review and
28. Wilcox CS. Regulation of renal blood flow by plasma chloride. J Clin Invest network meta-analysis. Ann Intern Med 2014;161:347-55.
1983;71:726-35. 34. Young P, Bailey M, Beasley R, Henderson S, Mackle D, McArthur C, et al.
29. Williams EL, Hildebrand KL, McCormick SA, Bedel MJ. The effect of in- Effect of a buffered crystalloid solution vs saline on acute kidney injury
travenous lactated Ringer’s solution versus 0.9% sodium chloride solu- among patients in the intensive care unit: the SPLIT randomized clinical
tion on serum osmolality in human volunteers. Anesth Analg 1999;88:999- trial. JAMA 2015;314:1701-10.
1003. 35. Ayus JC, Achinger SG, Arieff A. Brain cell volume regulation in hypona-
30. Kellum JA, Song M, Li J. Lactic and hydrochloric acids induce different tremia: role of sex, age, vasopressin, and hypoxia. Am J Physiol Renal Physiol
patterns of inflammatory response in LPS-stimulated RAW 264.7 cells. 2008;295:F619-24.
Am J Physiol Regul Integr Comp Physiol 2004;286:R686-92. 36. Balamuth F, Weiss SL, Hall M, Neuman MI, Scott H, Brady PW, et al.
31. Annane D, Siami S, Jaber S, Martin C, Elatrous S, Declere AD, et al. Effects Identifying pediatric severe sepsis and septic shock: accuracy of diagno-
of fluid resuscitation with colloids vs crystalloids on mortality in sis codes. J Pediatr 2015;167:1295-300.
Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis: A Matched Retrospective Cohort Study 7
Figure 1. Patient selection and matching. Patients receiving any LR (LR-any group) and patients receiving only LR (LR-only
group) were separately matched to patients receiving only NS (NS group). A suitable match from the NS group was identified
for 98.5% of patients who received any LR and 100% of patients who received only LR. DRG, diagnosis related group; NICU,
neonatal intensive care unit.
7.e1 Weiss et al
Figure 2. Summary of A, P values and B, standardized differences in unmatched and matched cohorts. The standardized dif-
ference is computed dividing the mean difference between matched patients by the pooled SD before matching, with values
<.10 considered an acceptable level of discrepancy within a matched pair. Before matching, there were statistically significant
differences in more than one-half of the covariates, and the standardized differences exceeded 0.10 in nearly one-quarter of
covariates. After matching, the median P value across covariates increased to >.25, and the standardized differences were <.10
for all covariates.
Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis: A Matched Retrospective Cohort Study 7.e2
Figure 5. Summary of A, P values and B, standardized differences across matched LR-any and NS patients within quartile of
total crystalloid volume. The standardized differences were <.10 for nearly all covariates in each quartile, reflecting an accept-
able maximum discrepancy within matched pairs in each volume quartile. However, the distribution of P values decreased with
each successive volume quartile, suggesting that the matching algorithm performed least effectively for those patients who re-
ceived the large total crystalloid fluid volumes.
Figure 6. Hospital mortality and AKI for patients matched after first stratifying LR-any group by proportion of total crystalloid
volume ordered as LR. Patients in the LR-any group were matched within each stratum to patients who received only NS. Each
set of bars is grouped by total crystalloid volume given as LR, including patients who only received NS but who were matched
to LR-any patients in those strata. Although both 30-day hospital mortality and AKI occurred less frequently as the proportion
of total fluids given as LR increased, there were no significant differences between the matched LR and NS groups within each
strata with the exception of significantly lower 30-day hospital mortality in the LR group within the 26% to ≤50% strata (*P = .03).
7.e3 Weiss et al
*Where 3- or 4-digit codes are listed, all associated subcodes were included.
Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis: A Matched Retrospective Cohort Study 7.e4
Table III. Premier codes for broad-spectrum antibiotics Table III. Continued
Premier standard Premier standard
Premier antibiotic names charge codes Premier antibiotic names charge codes
7.e5 Weiss et al
Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis: A Matched Retrospective Cohort Study 7.e6
Table III. Continued Table IV. Patient characteristics in full cohort before
Premier standard matching
Premier antibiotic names charge codes
LR group† NS group‡
Piperacillin/Tazo, Zosyn 3/0.375 g 250250052630000 Variables* n = 2150 n = 10 379 P value§
Piperacillin/Tazo, Zosyn 4/0.5 g 250250052640000
Piperacillin/Tazo, Zosyn 4/0.5 g 250250052860000 Age (y), median (IQR) 8 (1-15) 5 (1-13) <.001
Piperacillin/Tazo, Zosyn 3/0.375 g 250250052870000 Male sex 1136 (53) 5521 (53) .78
Piperacillin/Tazo, Zosyn 2/0.25 g 250250052890000 Race/Ethnicity .09
Quinupristin/Dalfopristin, Synercid 500 mg 10 mL 250250056410000 White 1004 (47) 4991 (48)
Ticar/Clav Pot, Timentin Adv 3.1 g 250250062680000 Black 456 (21) 2245 (22)
Ticar/Clav Pot, Timentin Intravenous Premix 3.1 g 250250062690000 Hispanic 252 (12) 1028 (10)
Ticar/Clav Pot, Timentin 3.1 g 250250062700000 Other 438 (20) 2115 (20)
Ticar/Clav Pot, Timentin 3.1 g 250250062710000 Comorbid conditions¶
Ticarcillin, Ticar Adv 3 g 250250062720000 Cardiovascular 371 (17) 1568 (15) .01
Ticarcillin, Ticar 3 g 250250062730000 Respiratory 79 (4) 505 (5) .02
Ticarcillin, Ticar 1 g 250250062740000 Renal 44 (2) 185 (2) .43
Ticarcillin, Ticar 20 g 250250062750000 Gastrointestinal 63 (3) 251 (2) .17
Ticarcillin, Ticar 30 g 250250062760000 Malignancy 191 (9) 1144 (11) <.001
Ticarcillin, Ticar 3 g 250250062770000 Hematologic/immunologic 121 (6) 671 (6) .16
Ticarcillin, Ticar 6 g 250250062780000
Metabolic 121 (6) 603 (6) .01
Tmp-Smz, Bactrim 10 mL 250250063080000
Neuromuscular 406 (19) 1366 (23) .08
Tmp-Smz, Bactrim 20 mL 250250063090000
Tmp-Smz, Bactrim 30 mL 250250063100000
Congenital disorders 211 (10) 923 (9) <.001
Tmp-Smz, Bactrim 50 mL 250250063110000 PICU admission 1761 (82) 7594 (73) <.001
Tmp-Smz, Bactrim 5 mL 250250063120000 Sepsis-related therapies**
Tmp-Smz, Bactrim Adv 10 mL 250250063130000 Vasoactive infusion 904 (42) 3416 (33) <.001
Tmp-Smz, Bactrim Adv 5 mL 250250063140000 Maximum number of 2 (1-3) 2 (1-2) .05
Tobramycin, Nebcin Inj 40 mg/mL 1.5 mL 250250063150000 concurrent vasoactives††
Tobramycin, Nebcin Inj 40 mg/mL 2 mL 250250063160000 Noninvasive mechanical 189 (9) 816 (8) .15
Tobramycin, Nebcin 100 mg 250250063170000 ventilation
Tobramycin, Nebcin 120 mg 250250063180000 Invasive mechanical ventilation 1419 (66) 5902 (57) <.001
Tobramycin, Nebcin 20 mg 250250063190000 Corticosteroids 567 (26) 2617 (25) .26
Tobramycin, Nebcin 40 mg 250250063200000 Hydrocortisone 235 (11) 1075 (10) .44
Tobramycin, Nebcin 60 mg 250250063210000 Methylprednisolone 404 (19) 1792 (17) .09
Tobramycin, Nebcin 80 mg 250250063220000 Albumin 5% 461 (21) 1690 (16) <.001
Tobramycin, Nebcin Pwdr 1.2 g 250250063230000 Albumin 25% 627 (29) 1981 (19) <.001
Tobramycin, Nebcin 10 mg/mL 2 mL 250250063240000 Blood transfusion‡‡ 1020 (47) 3653 (35) <.001
Tobramycin, Nebcin 10 mg/mL 6 mL 250250063250000
Furosemide 975 (45) 3846 (37) <.001
Tobramycin, Nebcin 10 mg/mL 8 mL 250250063260000
ECMO 12 (<1) 37 (<1) .18
Tobramycin, Nebcin 40 mg/mL 1 mL 250250063270000
Tobramycin, Nebcin 40 mg/mL 2 mL 250250063280000
ECMO, extracorporeal membrane oxygenation.
Tobramycin, Nebcin 40 mg/mL 30 mL 250250063290000 *Data presented as n (%), unless noted.
Trimetrexate, Neutrexin 25 mg 5 mL 250250065010000 †LR group included all patients who received any amount of LR fluid resuscitation.
Trovafloxacin, Trovan 5 mg/mL 40 mL 250250065370000 ‡NS group included patients who received only NS fluid resuscitation.
Trovafloxacin, Trovan 5 mg/mL 60 mL 250250065380000 §Statistical comparison using Wilcoxon signed rank and McNemar test for matched pairs, as
Trovafloxacin, Trovan 200 mg 250250065410000 appropriate.
Trovafloxacin, Trovan 300 mg 250250065420000 ¶Comorbid conditions were categorized using ICD-9-CM codes defining pediatric complex chronic
Vancomycin, Vancocin Adv 1 g 250250065800000 conditions.19
Vancomycin, Vancocin Adv 500 mg 250250065810000 **Includes therapies administered through hospital day 3.
Vancomycin, Vancocin 1 g 250250065840000 ††Includes only patients who received at least one vasoactive infusion.
Vancomycin, Vancocin 250 mg 250250065850000 ‡‡Includes administration of whole blood, packed red blood cells, platelets, plasma, and cryo-
Vancomycin, Vancocin 500 mg 250250065860000 precipitate.
Vancomycin, Vancocin 750 mg 250250065870000
Vancomycin, Vancocin 10 g 250250065930000
Vancomycin, Vancocin 1 g 250250065940000
Vancomycin, Vancocin 500 mg 250250065950000
Vancomycin, Vancocin 5 g 250250065960000
Linezolid, Zyvox 200 mg 250250073350000
Linezolid, Zyvox 400 mg 250250073360000
Linezolid, Zyvox 600 mg 250250073370000
Moxifloxacin, Avelox 400 mg 250250100010000
Ertapenem, Invanz 1 g 250250100400000
Daptomycin, Cubicin 250 mg 250250103650000
Daptomycin, Cubicin 500 mg 250250103660000
Levofloxacin, Levaquin 750 mg 250250103790000
Levofloxacin, Levaquin 750 mg 250250104050000
Levofloxacin, Levaquin 250 mg 250250108380000
Doripenem, Doribax 500 mg 250250108640000
Ceftaroline, Teflaro 400 mg 250250110450000
Ceftaroline, Teflaro 600 mg 250250110460000
Telavancin, Vibativ 250 mg 250250111670000
Telavancin, Vibativ 750 mg 250250111680000
7.e7 Weiss et al
Source: http://www.cdc.gov/growthcharts.
Table IX. Outcomes in matched cohort stratified by criteria used to identify severe sepsis/septic shock
Risk ratio/difference
Outcomes* LR-any group† NS group‡ (95% CI) P-value§
All patients with ICD-9-CM codes for severe sepsis or septic shock¶
Mortality, 30-d 57 (13.8) 68 (16.4) 0.97 (0.92, 1.03) .16
Mortality, hospital 66 (15.9) 78 (18.8) 0.97 (0.91, 1.03) .15
Mortality, including hospice 68 (16.4) 82 (19.8) 0.96 (0.90, 1.02) .12
AKI 125 (30.1) 122 (29.4) 1.0 (0.92, 1.10) .60
New dialysis 8 (1.9) 16 (3.9) 0.98 (0.96, 1.00) .07
PICU LOS**, median (IQR) 8.5 (2.0, 14.5) 8.6 (2.0,15.0) −0.05 (−1.2, 1.1) .95
Hospital LOS**, median (IQR) 14.8 (6.0, 22.0) 13.9 (5.0, 22.0) 0.88 (−0.6, 2.4) .34
Only patients with ICD-9-CM codes for infection and organ dysfunction
Mortality, 30-d 96 (5.6) 100 (5.9) 1.0 (1.47, 2.81) .41
Mortality, hospital 128 (7.5) 121 (7.1) 1.0 (0.99, 1.02) .71
Mortality, including hospice 132 (7.8) 129 (7.6) 1.0 (0.98, 1.02) .61
AKI 209 (12) 215 (13) 1.0 (0.97, 1.02) .40
New dialysis 19 (1.1) 17 (1.0) 1.0 (0.99, 1.01) .75
PICU LOS**, median (IQR) 7.5 (1.0, 13.0) 6.9 (1.0, 12.0) 0.6 (0.1, 1.1) .04
Hospital LOS**, median (IQR) 14.9 (6.0, 22.0) 12.8 (4.0, 20.0) 2.1 (1.5, 2.8) <.001
Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis: A Matched Retrospective Cohort Study 7.e8
Table X. Patient characteristics by quartile of total crystalloid volume corrected for estimated weight
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Variables* (n = 2816) (n = 3300) (n = 3212) (n = 3201) P-value†
Total crystalloid volume range, mL/kg 2 to 13 13.1 to 23.5 23.6 to 45.5 45.6 to 550
Age (y), median (IQR) 13 (8-16) 7 (1-14) 5 (1-13) 1 (0-4) <.001
Male sex 1434 (51) 1761 (53) 1740 (54) 1722 (54) .06
Race/ethnicity <.001
White 1470 (52) 1604 (49) 1490 (46) 1432 (45)
Black 587 (21) 68 (21) 691 (22) 735 (23)
Hispanic 249 (9) 348 (11) 354 (11) 329 (10)
Other 510 (18) 660 (20) 329 (10) 706 (22)
Comorbid conditions‡
Cardiovascular 272 (10) 430 (13) 496 (15) 741 (23) <.001
Respiratory 79 (3) 148 (4) 144 (4) 213 (7) <.001
Renal 49 (2) 59 (2) 65 (2) 56 (2) .81
Gastrointestinal 80 (3) 74 (2) 60 (2) 100 (3) .006
Malignancy 375 (13) 391 (12) 336 (10) 233 (7) <.001
Hematologic/Immunologic 237 (8) 218 (7) 194 (6) 143 (4) <.001
Metabolic 173 (6) 166 (5) 177 (6) 208 (7) .06
Neuromuscular 747 (27) 761 (23) 640 (20) 624 (19) <.001
Congenital disorders 300 (11) 286 (9) 276 (9) 272 (9) .01
PICU admission 1764 (63) 2322 (70) 2470 (77) 2799 (87) <.001
Sepsis-related therapies§
Vasoactive infusion 687 (24) 958 (29) 1115 (35) 1560 (48) <.001
Max number of concurrent vasoactives¶ 1 (1-2) 1 (1-2) 2 (1-2) 2 (1-3) <.001
Noninvasive mechanical ventilation 227 (8) 275 (8) 263 (8) 240 (8) .63
Invasive mechanical ventilation 1212 (43) 1721 (52) 1902 (59) 2486 (78) <.001
Corticosteroids 624 (22) 792 (24) 839 (26) 929 (29) <.001
Hydrocortisone 245 (9) 303 (9) 332 (10) 430 (13) <.001
Methylprednisolone 431 (15) 566 (17) 598 (19) 601 (19) .001
Albumin 5% 289 (10) 466 (14) 584 (19) 812 (25) <.001
Albumin 25% 335 (12) 542 (16) 676 (21) 1055 (33) <.001
Blood transfusion** 827 (29) 1089 (33) 1192 (37) 1565 (49) <.001
Furosemide 731 (26) 1096 (33) 1298 (40) 1696 (53) <.001
ECMO 7 (<1) 5 (<1) 10 (<1) 27 (<1) <.001
7.e9 Weiss et al
Table XI. Patient characteristics in LR-any group by proportion of total crystalloid fluids received as LR
1% to ≤25% LR 26% to ≤50% LR 51% to ≤75% LR 76% to <100% LR 100% LR
Variables* (n = 300) (n = 872) (n = 383) (n = 136) (n = 459) P-value†
Age (y), median (IQR) 11 (3-15) 10 (2-15) 8 (1-15) 5 (0-13) 5 (1-13) <.001
Male sex 168 (56) 463 (53) 192 (50) 82 (60) 231 (50) .16
Race/ethnicity .02
White 147 (49) 403 (46) 157 (41) 62 (46) 235 (51)
Black 58 (19) 186 (21) 82 (22) 31 (23) 98 (21)
Hispanic 21 (7) 102 (12) 56 (15) 19 (14) 54 (12)
Other 74 (25) 181 (21) 87 (23) 24 (18) 72 (16)
Comorbid conditions‡
Cardiovascular 56 (19) 156 (18) 61 (16) 26 (19) 72 (16) .69
Respiratory 7 (2) 31 (4) 10 (3) 8 (6) 23 (5) .14
Renal 6 (2) 14 (2) 10 (3) 3 (2) 11 (2) .78
Gastrointestinal 13 (4) 24 (3) 12 (3) 2 (1) 12 (3) .49
Malignancy 26 (9) 80 (9) 46 (12) 8 (6) 31 (7) .07
Hematologic/Immunologic 17 (6) 40 (5) 32 (8) 10 (7) 23 (5) .12
Metabolic 28 (9) 42 (5) 20 (5) 12 (9) 19 (4) .009
Neuromuscular 68 (23) 159 (18) 69 (18) 25 (18) 85 (19) .51
Congenital disorders 32 (11) 94 (11) 32 (8) 18 (13) 35 (8) .17
PICU admission 270 (90) 744 (85) 319 (83) 110 (81) 318 (69) <.001
Sepsis-related therapies§
Vasoactive infusion 166 (55) 417 (48) 164 (43) 53 (39) 104 (23) <.001
Max number of concurrent vasoactives¶ 2 (1-3) 2 (1-3) 1 (1-2) 1 (1-2) 1 (1-2) .004
Noninvasive mechanical ventilation 31 (10) 85 (10) 32 (8) 5 (4) 36 (8) .14
Invasive mechanical ventilation 232 (77) 606 (70) 247 (64) 87 (64) 247 (54) <.001
Corticosteroids 100 (33) 235 (27) 100 (26) 42 (31) 90 (20) .001
Hydrocortisone 54 918) 99 (11) 34 (9) 17 (13) 31 (7) <.001
Methylprednisolone 62 (21) 164 (19) 74 (19) 33 (24) 71 (15) .15
Albumin 5% 77 (26) 235 (27) 80 (21) 25 (18) 44 (10) <.001
Albumin 25% 89 (30) 279 (32) 133 (35) 40 (29) 86 (19) <.001
Blood transfusion** 160 (53) 459 (53) 208 (54) 65 (48) 128 (28) <.001
Furosemide 159 (53) 438 (50) 176 (46) 55 (40) 147 (32) <.001
ECMO 4 (1) 5 (1) 1 (<1) 0 2 (<1) .31
Table XII. Outcomes in LR-any group by proportion of total crystalloid fluids received as LR
Outcomes* 1% to ≤25% LR 26% to ≤50% LR 51% to ≤75% LR 76% to <100% LR 100% LR
n 300 872 383 136 459
Mortality, 30-d 38 (12.7) 67 (7.7) 20 (5.2) 12 (8.8) 23 (5.0)
Mortality, hospital 43 (14.3) 87 (10) 27 (7.0) 16 (11.8) 29 (6.3)
Mortality, including hospice 44 (14.7) 90 (10.3) 28 (7.3) 16 (11.8) 30 (6.5)
AKI 75 (25.0) 145 (16.7) 57 (14.9) 18 (13.2) 45 (9.8)
New dialysis 9 (2.5) 9 (1.0) 3 (<1) 2 (1.5) 5 (1.1)
PICU LOS†, median (IQR) 9.7 (0, 7.0) 8.2 (1.0, 12.0) 7.1 (1, 11.0) 7.7 (3.0, 16.0) 5.8 (0, 9.0)
Hospital LOS†, median (IQR) 15.6 (4.0, 14.0) 16.1 (5.0, 21.0) 16.0 (5.0, 19.0) 14.9 (7.0, 23.0) 11.8 (4, 17)
Crystalloid Fluid Choice and Clinical Outcomes in Pediatric Sepsis: A Matched Retrospective Cohort Study 7.e10