Hypotonic versus Isotonic Fluids in Hospitalized Children: A SystematicReview and Meta-Analysis

Byron Alexander Foster, MD, MPH1, Dina Tom, MD1, and Vanessa Hill, MD2

Objective To determine whether the use of hypotonic vs isotonic maintenance fluids confers an increased risk ofhyponatremia in hospitalized children.Study design A search of MEDLINE (1946 to January 2013), the Cochrane Central Registry (1991 to December2012), Cumulative Index for Nursing and Allied Health Literature (1990 to December 2012), and Pediatric AcademicSocieties (2000-2012) abstracts was conducted using the terms “hypotonic fluids/saline/solutions” and “isotonicfluids/saline/solutions,” and citations were reviewed using a predefined protocol. Data on the primary and second-ary outcomes were extracted from original articles by 2 authors independently. Meta-analyses of the primary andsecondary outcomes were performed when possible.Results A total of 1634 citations were screened. Ten studies (n = 893) identified as independent randomizedcontrolled trials were included. Five studies examined subjects in the intensive care unit setting, including 4 on reg-ular wards and 1 in amixed setting. In hospitalized children receivingmaintenance intravenous fluids, hyponatremiawas seenmore often in those receiving hypotonic fluids than in those receiving isotonic fluids, with an overall relativerisk of 2.37 (95% CI, 1.72-3.26). Receipt of hypotonic fluids was associated with a relative risk of moderate hypo-natremia (<130 mmol/L) of 6.1 (95% CI, 2.2-17.3). A subgroup analysis of hypotonic fluids with half-normal salinefound a relative risk of hyponatremia of 2.42 (95% CI, 1.32-4.45).Conclusion In hospitalized children in intensive care and postoperative settings, the administration of hypotonicmaintenance fluids increases the risk of hyponatremia when compared with administration of isotonic fluids. Forpatients on general wards, insufficient data are available based on the reviewed studies, and individual risk factorsmust be assessed. (J Pediatr 2014;-:---).

See editorial, p ���

pproximately 2.2 million non-newborn children are admitted to the hospital every year in the US, many of whom

Areceive maintenance intravenous (IV) fluids during their stay.1,2 There is a long tradition of prescribing precise main-tenance fluid therapy to provide appropriate amounts of water, glucose, and electrolytes for hospitalized children.3,4

The volume of maintenance fluid is calculated using an adaptation of Holliday and Segar’s work,4 now known as the “4-2-1rule.” This calculation relies on the relationship between energy and body weight to calculate water needs. In addition to calcu-lating the amount of water needed, Holliday and Segar created the maintenance fluid in a hypotonic solution to best approx-imate solute needs. However, we now understand that water regulation in sick children is affected by both osmotic andnonosmotic (ie, stress, infection) stimuli for increased antidiuretic hormone (ADH) production.5,6

Given our current understanding of fluid regulation, the composition of those fluids has been the subject of debate,7-9 partic-ularly regarding tonicity. The UK Royal College of Pediatrics has questioned the use of hypotonic fluids, issuing safety warningsregarding their use.10 Nevertheless, a study showed that 78% of pediatric residents prescribed hypotonic fluids in hypotheticalclinical situations inwhich the patient was at high risk for increasedADH secretion,11 and thus at increased risk for hyponatremia.

When administering hypotonic fluids, a subset of children have a small but significant individual risk of hyponatremia andits sequelae, including brain swelling and death.12 Given the extent of IV fluid use, calculating a population estimate of risk from

From the 1Division of Inpatient Pediatrics, Department ofPediatrics, University of Texas Health Science Center atSan Antonio and 2Department of Pediatrics, BaylorCollege of Medicine, San Antonio, TX

B.F. is a scholar of the Research to Advance CommunityHealth Center. The authors declare no conflicts ofinterest.

this small individual risk allows estimation of the potential impact of this risk.A meta-analysis by Choong et al13 found a significant risk of developing hypo-

natremia using hypotonic fluids compared with isotonic fluids (OR, 17.22; 95%CI, 8.67-34.2). They showed consistency over multiple outcomes; however, theirmeta-analysis included both randomized and cohort studies. Multiple random-ized, clinical trials have followed, primarily in postoperative and intensive careunit (ICU) patients.

Portions of this study were presented as an abstract,which received the 2013 Abstract Research Award fromthe American Academy of Pediatrics, Section of HospitalMedicine at the AAP National Conference & Exhibition inOrlando, FL, October 26-29, 2013.

0022-3476/$ - see front matter. Copyright ª 2014 Mosby Inc.

All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2014.01.040

ADH Antidiuretic hormone

ICU Intensive care unit

IV Intravenous

1

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. -, No. -

In the present study, we aimed to address the question ofwhether hypotonic saline vs isotonic saline infused at main-tenance rates in hospitalized children confers an increasedrisk of developing hyponatremia, defined here as a serum so-dium level <135 mmol/L.

Methods

We developed a detailed protocol for the selection of studies(available on request from the authors). Studies eligible for in-clusion were clinical trials that enrolled hospitalized childrenaged1month to18 years and compared isotonic andhypotonicIV fluids. Studies were excluded if they did not have a compar-ison grouporonly studiedpatientswithdiabetes insipidus, dia-betic ketoacidosis, burns, or shock, because the former isknown to involve sodiumdysregulation and the latter 3 requiredisease-specific protocols for fluid management.

A search of MEDLINE (1946 to January 2013), CochraneCentral Registry (1991 to December 2012), CIANHL (1990to December 2012), and PAS (2000-2012) abstracts wascompleted in January 2013. Terms used in the search were“hypotonic fluids/saline/solutions” and “isotonic fluids/sa-line/solutions,” with the restrictions of children and clinicaltrial. The results from the initial search were cross-referencedand liberally screened by title. For the second round, 2 of theauthors independently screened abstracts and full articles, asnecessary, using the inclusion and exclusion criteria. Authorsof unpublished or partially published studies that resultedfrom the search were contacted for further information. Twoof the authors conducted independent hand searches of thereference lists of all selected full text articles. Only randomizedcontrolled trials were included in the meta-analysis.

Our primary outcome was hyponatremia, defined as aserum sodium level <135 mmol/L with assessment at thelongest time point for each study that did not exceed24 hours. Secondary outcomes were change in serum sodiumlevel from baseline, characterized as moderate hyponatremia(serum sodium <130 mmol/L) or severe hyponatremia(serum sodium <125 mmol/L). Adverse events of hyperna-tremia (serum sodium >145 mmol/L); edema; hypertension;neurologic complications, defined as new-onset systemicneurologic symptoms (eg, seizures, altered mental status);and mortality were assessed as well.

We collected data using a standardized form that was pilottested by 2 of the authors. The authors of primary studies werecontacted for clarification of results. Data extracted for eachstudy included study date and duration; number and age ofsubjects enrolled and completed in each group; interventionscompared, including fluid infusion rates and tonicity; setting(ICU vs ward); postoperative (surgical) or medical status;exclusion criteria and comorbidities of subjects; outcomes as-sessed and time points of data collection; and the primary andsecondary outcomes as defined by this review. Hypotonicfluids were defined as tonicity <250 mmol/L for this review.Isotonic fluids were defined as normal (0.9%) saline, Ringerlactate, Hartmann solution, and any other fluid with tonicityapproaching that of normal serum.

2

Using the criteria proposed by the Cochrane group,14 weassessed each study in the final analysis for random sequencegeneration, allocation concealment, blinding of participantsand personnel, blinding of outcome assessment, incompleteoutcome data, and selective reporting. Two authors indepen-dently graded these variables for each individual study as low,high, or unclear level of bias. An article with a high risk of biaswas evaluated by reviewers for inclusion status in the finalanalysis. We assessed the risk of bias across studies, specif-ically publication bias, was assessed using a funnel plotwith RevMan 5.2 (The Nordic Cochrane Centre, The Co-chrane Collaboration, Copenhagen, Denmark).15

We conducted a brief qualitative review of the contributionof the cohort studies excluded from themeta-analysis (Table I;available at www.jpeds.com). The purpose of this reviewwas todeterminewhether outcomes differed significantly between thenonrandomized studies and the randomized clinical trials. Twoauthors extracted data independently as described above. Foreach study, we examined the study design and setting, subjectcharacteristics, type of intervention or comparison, andoutcome data related to serum sodium concentration.

Statistical AnalysesWe used a random-effects model to report relative risk for theindividual studies and the overall estimate of risk in the meta-analysis. We applied Mantel-Haenzel methods to estimate thecombined relative risk estimate for dichotomous outcomesand inverse variance for the continuous outcome of changein serum sodium concentration using RevMan 5.2.15 I2 statis-tics were used to assess heterogeneity. We used the followingequation to calculate the number needed to harm:

number needed to treat ¼ 1=½assumed control event rate

� ð1� relative risk estimateÞ�:14

We used a range of values (5%-20%) from the publishedliterature for the assumed control event rate of the primaryoutcome of hyponatremia.Based on theoretical concerns regarding applicability be-

tween groups, we a priori defined subgroups for sensitivityanalysis as postoperative surgical subjects vs nonoperative(ie, medical) subjects. After reviewing the trials for inclusion,we conducted a secondary sensitivity analysis of the risk forhyponatremia (ie, sodium<135mmol/L) in patients receiving1/2 normal saline (0.45%) vs those receiving isotonic fluid.The rationale for this approach was to examine more currentpractice, given that use of 0.5 normal 0.45% saline has becomeaccepted practice,11 and the amount of free water in 0.5normal 0.45% saline is one-half that of 0.25 normal 0.225%saline, which can significantly affect the intervention.

Results

Our initial search identified 1634 articles, which, aftercross-reference and initial screening, we narrowed to 85

Foster, Tom, and Hill

Table II. Characteristics of included studies

StudySubject

characteristicsInclusion criteria

and settingInterventions compared

(all with dextrose unless noted)Outcome data

(Na <135 mmol/L)

Coulthard et al 201216 n = 79; 4-14 y PICU, postoperative(spinal or craniotomy)

Hartmann’s vs 0.45% saline 0 of 39 (0%) in isotonic;7 of 40 (18%) in hypotonic

Rey et al 201117 n = 84; 2-10 y PICU, mixed postoperative(45%) and medical

156 mmoL/L tonicity vs 50-70 mmoL/L 8 of 45 (17.8%) in isotonic;19 of 39 (48.7%) in hypotonic

Yung and Keeley 200918 n = 61; 30 d-18 y PICU, mixed postoperativeand medical

0.9% saline vs 0.18% saline 3 of 29 (10.3%) in isotonic;7 of 32 (21.9%) in hypotonic

Monta~nana et al 200819 n = 122; 29 d-18 y PICU, postoperative in 84%(thoracic, cardiac,abdominal, CNS)

Na 140 mmol/L, K 15 mmol/L tonicityvs 20-100 mEq/L

3 of 59 (5.1%) in isotonic; 13 of63 (20.6%) in hypotonic

Saba et al 201120 n = 37; 3 mo-18 y Floor, mixed postoperative(67%) and medical

0.9% saline vs 0.45% saline 1 of 16 (6%) in isotonic; 1 of21 (5%) in hypotonic

Choong et al 201121 n = 218; 6 mo-16 y Mixed floor and PICU,postoperative

0.9% saline vs 0.45% saline, withor without KCl

26 of 106 (24.5%) in isotonic;47 of 112 (42%) in hypotonic

Neville et al 200622 n = 42; 6 mo-14 y Floor, gastroenteritis 0.9% saline vs 0.45% saline 0 of 20 (0%) in isotonic; 5 of 22(22.7%) in hypotonic

Brazel and McPhee 199623 n = 12; 12-18 y PICU, postoperative Hartmann’s vs 0.3% saline or0.18% saline

1 of 5 (20%) in isotonic; 7 of 7(100%) in hypotonic

Neville et al 201024 n = 124; 6 mo-15 y Floor, postoperative 0.9% saline vs 0.45% saline 1 of 31 (3%) in isotonic; 9 of31 (29%) in hypotonic

Kannan et al 201025 n = 114; 3 mo-12 y Floor, medical 0.9% saline vs 0.18% saline 5 of 58 (8.6%) in isotonic;13 of 56 (23%) in hypotonic

CNS, central nervous system; PICU, pediatric ICU.

- 2014 ORIGINAL ARTICLES

abstracts and full text articles. After final review and discus-sion, we selected 22 full text articles (Figure 1; available atwww.jpeds.com). Eleven articles were identified asrandomized controlled trials from which data wereextracted (Table II).16-26 One article was excludedbecause complete results were not available, given thatthe article was in the process of publication.26 The other11 articles were excluded because of study designfactors.27-37

Risk of BiasOur assessment of risk of bias within each individual studyrevealed the greatest risk of bias in the areas of attritionand blinding (Table III). The attrition bias was relatedprimarily to no outcome (sodium measurement)performed in subjects with no clear explanation. Althoughthere was no clear evidence of an attrition bias in eitherdirection, the risk remained. In the studies with a high riskof blinding of participants and personnel, it is unlikely thatthis had any significant impact on the outcome serumsodium concentration.

Table III. Risk of bias in included studies

StudyRandom sequence

generationAllocation

concealmentBlinding

and

Coulthard et al 201216 Low LowRey et al 201117 Low UnclearYung and Keeley 200918 Low LowMonta~nana et al 200819 Low LowSaba et al 201120 Low LowChoong et al 201121 Low LowNeville et al 200622 Unclear LowBrazel and McPhee199623 Unclear UnclearNeville et al 201024 Unclear LowKannan et al 201025 Low Low

Hypotonic versus Isotonic Fluids in Hospitalized Children: A Syst

The combination of poor blinding and incompleteoutcome data does pose an actual risk of bias, which ispossible in both previous reports.22,24 The risk of bias acrossstudies related to this subject area was assessed using a funnelplot, which was found to be not skewed (data not shown).

Primary OutcomeWe found an overall relative risk of hyponatremia, defined as aserum sodium concentration <135 mmol/L, of 2.37 (95% CI,1.72-3.26) in the subjects who received hypotonic fluidscompared with those who received isotonic fluids (Figure 2,A). Using an estimated control event rate of hyponatremia of5% as the lower range of published data in the literature,16-25

we obtained a number needed to harm of 15 (95% CI, 9-28).For a higher-risk population using the higher publishedevent rate of hyponatremia of 20% from the literature,16-25

the number needed to harm was 4 (95% CI, 3-7). Threestudies performed multiple comparisons of IV fluids usingalternate rates18,24,25; in those cases, we used the comparisonarm of the maintenance rate for the aforementioned analyses.

of participantspersonnel

Blinding ofoutcome

Incompleteoutcome data Selective reporting

Low Low Low LowUnclear Low Low LowLow Low Low Low

Unclear Low Low LowLow Low High LowLow Low Low LowHigh Low High Low

Unclear Low Low LowHigh Low High LowHigh Low Low Low

ematic Review and Meta-Analysis 3

Figure 2. A, Forest plot of all included randomized controlled trials using a random-effects model and M-H statistics with theoutcome of hyponatremia (sodium <135 mmol/L). B, Forest plot of randomized controlled trials using a random-effects modeland inverse variance to estimate mean difference in sodium (in mEq/L) between groups. M-H, Mantel-Haenzel.

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. -, No. -

Secondary OutcomesFor the secondary outcome of change in serum sodium con-centration, we were able to collect data on 5 studies, primarilyvia correspondence with authors.16,18,20,22,24 In those 5studies, the time point for assessment of change of serum so-dium ranged from 4 hours to 24 hours. We found amean dif-ference in the change in serum sodium from a baseline of�2.46 (95% CI, �3.11 to �1.81), I2 = 20%, in subjectsreceiving hypotonic fluids compared with those receivingisotonic fluids (Figure 2, B).

Mortality was not reported in 9 studies, presumablybecause it did not occur. In 1 study,25 1 death was reportedin the isotonic group secondary to acute respiratory distresssyndrome; that patient had a normal serum sodium levelthroughout.

Moderate hyponatremia (<130 mmol/L) was reported in 8studies, with 2 studies18,22 not reporting any data (Figure 3,A). The relative risk of moderate hyponatremia usinghypotonic fluids vs isotonic fluids was 6.1 (95% CI, 2.2-17.3). Severe hyponatremia (<125 mmol/L) was specificallyexamined in 6 studies.16,19,20,23-25 In those studies,encompassing 488 subjects, there were 3 cases of severehyponatremia,23,25 with 2 patients receiving hypotonicfluids and 1 patient with an unclear fluid group assignment.25

Neurologic impairment was described as an assessedoutcome in 4 of the 10 studies evaluated. Among thesestudies including at least 430 subjects,18,20,21,25 1 neurologicevent was reported in a patient with porphyria who had astarting serum sodium level of 132 mmol/L, received hypo-tonic fluids, and developed encephalopathy, seizures, and aserum sodium nadir of 124 mmol/L.25

Edema and hypertension were reported as secondary out-comes in 4 studies.19-21,25 In 1 study, 7 of 59 (11.9%) subjects

4

receiving isotonic fluids developed hypertension, comparedwith 7 of 63 (11.1%) subjects receiving hypotonic fluids.19

Another study described new-onset hypertension in 0 of128 (0%) of isotonic fluid recipients vs 2 of 130 (1.5%) hypo-tonic fluid recipients.21 The same study found generalizededema in 1 of 128 (0.8%) isotonic fluid recipients and in 7of 130 (5.4%) hypotonic fluid recipients.21 Two studiesthat reported assessing 1 or both of these outcomes reportedno events for either outcome.20,25

Hypernatremia (serum sodium >145 mmol/L) was as-sessed as an outcome in 8 studies.16,17,19-22,24,25 The relativerisk of hypernatremia using hypotonic fluids vs isotonicfluids was found to be 0.81 (95% CI, 0.32-2.04) (Figure 3,B). A range of 0-6% for incidence of hypernatremia wasreported, the number needed to harm using isotonic fluidswas not calculated given the nonsignificant findings.

Sensitivity Analyses and Subgroup AnalysesA sensitivity analysis excluding the 2 studies with possiblebias22,24 described above found similar results similar to theoverall risk seen in all studies (relative risk, 2.17; 95% CI,1.61-2.91; I2 = 0%). For the prespecified subgroup analysisof postoperative vs nonoperative patients, we were unableto complete the analysis because of lack of informationprovided in the studies of mixed patients. For the posthocsubgroup analysis of hypotonic fluids using at least half-normal saline, we found similar results to the overall analysiswith a relative risk of 2.42 (95% CI, 1.32-4.45).

Discussion

We found a significantly increased risk of hyponatremia inhospitalized patients receiving hypotonic fluids at

Foster, Tom, and Hill

Figure 3. A, Forest plot of randomized controlled trials using a random-effects model and M-H statistics for the secondaryoutcome of moderate hyponatremia (serum sodium <130 mmol/L). B, Forest plot of randomized controlled trials using arandom-effects model and Mantel-Haenzel statistics for the secondary outcome of hypernatremia (Na >145 mmol/L).

- 2014 ORIGINAL ARTICLES

maintenance rates. This risk was consistent looking at changein serum sodium and moderate hyponatremia, as well as inthe subset of trials using 0.5 normal 0.45% saline as the com-parison fluid. This review adds to the previous systematic re-view13 by including 8 additional randomized controlled trialswith 777 additional subjects; the majority of these subjectswere in an ICU or postoperative setting.

Based on previous hypotheses suggesting that hypotonicfluids can be used safely but at lower thanmaintenance rates,4

we examined 2 studies that compared hypotonic fluids atone-half or two-thirds the maintenance rate. Neville et al24

found a similar rate of hyponatremia in a group receivinghypotonic fluids at maintenance rate and a group receivinghypotonic fluids at one-half maintenance rate (29% vs32%). In the study of Yung and Keeley,18 fluid type, notrate, was identified as the sole significant predictor of hypo-natremia. This would seem to suggest that any source of freewater is sufficient in the situation of ADH excess.

Themean change in serum sodium concentration of�2.46(95% CI,�3.11 to�1.81) using hypotonic fluids likely is notclinically significant. One of the controversies around whichthe types of fluids to use routinely in hospitalized children re-volves around the provision of maintenance vs replacementfluids.7 Most of the patients studied in this review, and allthe studies included in the subanalysis of 0.5 normal 0.45%saline, were judged to be euvolemic.

In this study, we were not able to separate out the opera-tive, surgical subjects from the nonoperative subjects. Basedon calculations of described subjects, at least 598 of the 893subjects (67%) were postoperative surgical patients. It is

Hypotonic versus Isotonic Fluids in Hospitalized Children: A Syst

not possible to extend the conclusions from this analysis toall hospitalized children. Hyponatremia has been describedin common hospital pediatric conditions, such as bronchio-litis38 and pneumonia,39 with an incidence of moderate hy-ponatremia of 8%-10%, similar to the range of 6%-14%reported in the studies reviewed here.The study of Saba et al20 stands as an outlier. That studywas

powered to assess a change in serum sodium concentration,with a population similar to that of other studies with themajority of subjects were postoperative surgical or had gastro-enteritis. Inmany of these studies, the incidence of hyponatre-mia was approximately 20% for the hypotonic fluid arm,whereas Saba et al reported rates of 5% for the hypotonic fluidarm and 6% for the isotonic fluid arm, possibly explaining thedifference in findings. The reason for the difference in inci-dence of hyponatremia is not entirely clear, however.Finally, one-half of the studies that we evaluated were con-

ducted in the ICU setting. The ICU setting may be a proxy fora degree of physiological stress under which free water is re-tained to a greater extent; however, an argument against thisis the consistency of results between studies conducted in theICU and those conducted on the regular floor (Table II).Choong et al21 studied a mixed ICU and floor populationand found no association with ICU status andhyponatremia after taking the fluid type into account.Limitations of the present analysis include a lack of studies

assessing nonoperative patients receiving 0.5 normal 0.45%saline compared with isotonic fluids. The sole study that hasexamined such a population arguably included hypovolemicchildren and used a nonmaintenance protocol.22 Some studies

ematic Review and Meta-Analysis 5

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. -, No. -

reported a potential for bias, primarily around attrition, withsodium levels not collected in a subset of patients, and somepotential for bias in blinding. Some studies allowed for oralintake, and some used oral intake as a marker for discontinu-ation of fluids. Most of the time points were 24 hours in thisreview; a longer follow-up might have revealed either higherrates of hyponatremia or physiological equilibrium, but thissubanalysis was not possible. There is a potential bias forthe outcome of change in sodium concentration, consideringthat these data were received from only 5 authors.

The included studies were not powered to detect a differ-ence in adverse effects of hypertension, neurologic sequelae,or hypernatremia. We found a significant risk of moderatehyponatremia in the subjects receiving hypotonic fluids;few studies reported severe hyponatremia or neurologicsequelae. Studying this may prove difficult given the largesample size required and the low incidence of neurologicsequelae,12 and intervening before that threshold is reachedarguably should be done in a research study.

Recent studies have improved our understanding of thepossible mechanisms underlying the nonosmotic stimulifor ADH excess.40,41 Many hospitalized children experiencerisk factors for nonosmotic stimulation of ADH. The chal-lenge lies in determining which patients are at risk for hypo-natremia, for which no reliable assessment tool is currentlyavailable besides clinical experience and judgment. The evi-dence reviewed here demonstrates an increased risk of hypo-natremia, including moderate hyponatremia, in childrenempirically prescribed a source of free water, namely hypo-tonic fluids. These conclusions are limited to the postopera-tive and ICU populations, however, given the sparse data ongeneral ward patients, and assessment of individual riskshould continue. n

We would like to thank KoKo Aung, MD, MPH, for his advice on themeta-analysis.

Submitted for publication Oct 31, 2013; last revision received Dec 23, 2013;

accepted Jan 21, 2014.

Reprint requests: Byron Alexander Foster, MD, MPH, Division of Inpatient

Pediatrics, Department of Pediatrics, University of Texas Health Science

Center at San Antonio, 7703 Floyd Curl Dr, MSC 7802, San Antonio, TX 78229.

E-mail: fosterba@uthscsa.edu

References

1. Yu H, Wier LM, Elixhauser A. Hospital stays for children, 2009. Health-

care Cost andUtilization Project statistical brief 118, August 2011. Avail-

able from: http://www.hcup-us.ahrq.gov/reports/statbriefs/sb118.pdf.

Accessed May 1, 2013.

2. HCUPnet. Healthcare Cost and Utilization Project (HCUP), 2006-2009.

Available from: http://hcupnet.ahrq.gov/. Accessed May 5, 2013.

3. Talbot NB, Crawford JD, Butler AM. Homeostatic limits to safe paren-

teral fluid therapy. N Engl J Med 1953;248:1100-8.

4. Holliday MA, Segar WE. The maintenance need for water in parenteral

fluid therapy. Pediatrics 1957;19:823.

5. GerigkM,GnehmHE,RascherW.Arginine vasopressin and renin in acutely

ill children: implication for fluid therapy. Acta Paediatr 1996;85:550-3.

6. Moritz ML, Ayus JC. Hospital-acquired hyponatremia: why are hypo-

tonic parenteral fluids still being used? Nat Clin Pract Nephrol 2007;3:

374-82.

6

7. Holliday MA, Ray PE, Friedman AL. Fluid therapy for children: facts,

fashions and questions. Arch Dis Child 2007;92:546-50.

8. Moritz ML, Ayus JC. Hospital-acquired hyponatremia: why are there

still deaths? Pediatrics 2004;113:1395-6.

9. Moritz ML, Ayus JC. Prevention of hospital-acquired hyponatremia: a

case for using isotonic saline. Pediatrics 2003;111:227-30.

10. National Patient Safety Agency alert: reducing the risk of hyponatraemia

when administering intravenous infusions to children. Available from:

http://www.nrls.npsa.nhs.uk/resources/?entryid45=59809&q=0%c2%ac

hypotonic%c2%ac. Accessed April 2013.

11. Freeman MA, Ayus JC, Moritz ML. Maintenance intravenous fluid pre-

scribing practices among paediatric residents. Acta Paediatr 2012;101:

e465-8.

12. Arieff AI, Ayus JC, Fraser CL. Hyponatraemia and death or permanent

brain damage in healthy children. BMJ 1992;304:1218-22.

13. Choong K, KhoME,MenonK, BohnD.Hypotonic versus isotonic saline

in hospitalized children: a systematic review. Arch Dis Child 2006;91:

828-35.

14. Higgins JPT, Green S, eds. Cochrane handbook for systematic reviews of

interventions, version 5.1.0. Available from www.cochrane-handbook.

org. Accessed April 2013.

15. RevMan version 5.2. Copenhagen, Denmark: Nordic Cochrane Centre,

the Cochrane Collaboration; 2012.

16. Coulthard MG, Long DA, Ullman AJ, Ware RS. A randomised

controlled trial of Hartmann’s solution versus half normal saline in post-

operative paediatric spinal instrumentation and craniotomy patients.

Arch Dis Child 2012;97:491-6.

17. Rey C, Los-Arcos M, Hern�andez A, S�anchez A, D�ıaz JJ, L�opez-Herce J.

Hypotonic versus isotonic maintenance fluids in critically ill children:

a multicenter prospective randomized study. Acta Paediatr 2011;100:

1138-43.

18. Yung M, Keeley S. Randomised controlled trial of intravenous mainte-

nance fluids. J Paediatr Child Health 2009;45:9-14.

19. Monta~nana PA, Modesto i Alapont V, Oc�on AP, L�opez PO, L�opez

Prats JL, Toledo Parre~no JD. The use of isotonic fluid as maintenance

therapy prevents iatrogenic hyponatremia in pediatrics: a randomized,

controlled open study. Pediatr Crit Care Med 2009;9:589-97.

20. Saba TG, Fairbairn J, Houghton F, Laforte D, Foster BJ. A randomized

controlled trial of isotonic versus hypotonic maintenance intravenous

fluids in hospitalized children. BMC Pediatr 2011;11:82.

21. Choong K, Arora S, Cheng J, Farrokhyar F, Reddy D, Thabane L, et al.

Hypotonic versus isotonic maintenance fluids after surgery for children:

a randomized controlled trial. Pediatrics 2011;128:857-66.

22. Neville KA, Verge CF, Rosenberg AR, O’Meara MW,Walker JL. Isotonic

is better than hypotonic saline for intravenous rehydration of children

with gastroenteritis: a prospective randomised study. Arch Dis Child

2006;91:226-32.

23. Brazel PW, McPhee IB. Inappropriate secretion of antidiuretic hormone

in postoperative scoliosis patients: the role of fluid management. Spine

1996;21:724-7.

24. Neville KA, Sandeman DJ, Rubinstein A, Henry GM, McGlynn M,

Walker JL. Prevention of hyponatremia duringmaintenance intravenous

fluid administration: a prospective randomized study of fluid type versus

fluid rate. J Pediatr 2010;156:313-9.

25. Kannan L, Lodha R, Vivekanandhan S, Bagga A, Kabra SK, Kabra M.

Intravenous fluid regimen and hyponatraemia among children: a ran-

domized controlled trial. Pediatr Nephrol 2010;252:303-9.

26. Pemde HK, Dutta AK, Sodani R, Mishra K. A randomized controlled

trial of intravenous maintenance fluids (isotonic versus hypotonic) in

young children with central nervous system infections. Pediatric Aca-

demic Societies abstracts archive: E-PAS2012:2922.402. Available from:

http://www.abstracts2view.com/pasall/. Accessed January 12, 2013.

27. Burrows FA, Shutack JG, Crone RK. Inappropriate secretion of antidi-

uretic hormone in a postsurgical pediatric population. Crit Care Med

1983;11:527-31.

28. Wilkinson E, Rieff J, Rekate HL, Beals S. Fluid, blood, and blood

product management in the craniofacial patient. Pediatr Neurosurg

1992;18:48-52.

Foster, Tom, and Hill

- 2014 ORIGINAL ARTICLES

29. Dagli G, OrhanME, Kurt E. The effects of different hydration fluids used

in pediatric anaesthesia on blood glucose, electrolytes, and cardiovascu-

lar stability. GATA B€ulteni 1997;39:146-52.30. Hoorn EJ, Geary D, RobbM. Acute hyponatremia related to intravenous

fluid administration in hospitalized children: an observational study. Pe-

diatrics 2004;113:1279-84.

31. Armon K, Riordan A, Playfor S, Millman G, Khader A. Paediatric

Research Society. Hyponatraemia and hypokalaemia during intravenous

fluid administration. Arch Dis Child 2008;93:285-7.

32. Au AK, Ray PE, McBryde KD, Newman KD,Weinstein SL, Bell MJ. Inci-

dence of postoperative hyponatremia and complications in critically ill

children treated with hypotonic and normotonic solutions. J Pediatr

2008;152:33-8.

33. Bar�on FJ, Balladores C, Carretero P, Lombardo L, Mannino L, Martin S.

Effects on natremia of the intravenous hypotonic fluid administration in

hospitalized children with acute lower respiratory tract infection. Arch

Argent Pediatr 2009;107:335-9.

34. Eulmesekian PG, P�erez A, Minces PG, Bohn D. Hospital-ac-

quired hyponatremia in postoperative pediatric patients: pro-

spective observational study. Pediatr Crit Care Med 2010;11:

479-83.

Hypotonic versus Isotonic Fluids in Hospitalized Children: A Syst

35. Hanna M, Saberi MS. Incidence of hyponatremia in children with

gastroenteritis treated with hypotonic intravenous fluids. Pediatr Neph-

rol 2010;25:1471-5.

36. Cladis FP, Bykowski M, Schmitt E, Naran S, Moritz ML, Cray J, et al.

Postoperative hyponatremia following calvarial vault remodeling in cra-

niosynostosis. Paediatr Anaesth 2011;21:1020-5.

37. Kanda K, Nozu K, Kaito H, Iijima K, Nakanishi K, Yoshikawa N, et al.

The relationship between arginine vasopressin levels and hyponatremia

following a percutaneous renal biopsy in children receiving hypotonic or

isotonic intravenous fluids. Pediatr Nephrol 2011;26:99-104.

38. Hanna S, Tibby SM, Durward A, Murdoch IA. Incidence of hyponatrae-

mia and hyponatraemic seizures in severe respiratory syncytial virus

bronchiolitis. Acta Paediatr 2003;92:430-4.

39. Don M, Valerio G, Korppi M, Canciani M. Hyponatremia in pediatric

community-acquired pneumonia. Pediatr Nephrol 2008;23:2247-53.

40. Ellison DH, Berl T. Clinical practice: the syndrome of inappropriate

antidiuresis. N Engl J Med 2007;356:2064-72.

41. Ferreira da Cunha D, Pontes Monteiro J, Modesto dos Santos V, Ara�ujoOliveira F, Freire de Carvalho da Cunha S. Hyponatremia in acute-phase

response syndrome patients in general surgical wards. Am J Nephrol

2000;20:37-41.

ematic Review and Meta-Analysis 7

Table I. Characteristics of relevant nonrandomized studies

Study Study design

Participant characteristics:age, n, setting, andtype of subjects

Interventions compared(dextrose in fluidsexcept as noted) Outcome data

Burrows et al 198327 Prospective cohort <18 y, 24, postoperativespinal fusion

0.225% saline or0.45% saline vs LR

Hyponatremia in 5 of20 in the hypotonicgroup vs 0 of 4 in theisotonic group.

Wilkinson et al 199228 Retrospective cohort 2 mo-14 y, 56, postoperativecraniofacial

Hypotonic vs isotonicfluids (multiple types)

D5 1 of 5NS: 9 of 18patients hadNa #130 mmol/L;D5 1/2NS: 1 of 6had Na <135 mmol/L;D5 LR: 2 of 15 had Na<135 mmol/L.

Dagli et al 199729 Clinical trial 1-12 y, 60, elective surgery(primarily intraoperative)

LR vs D1 LR vs D3.3with 0.3% saline

At 60 min postoperation,only 1/3NS grouphad a significant dropin Na; mean, 133 mmol/L;P < .05 for change frominitial level.

Hoorn et al 200430 Case-control 1-13 y, 148, admitted throughemergency room

IV fluid compositionbetween cases (n = 37)and controls (n = 111)

Greater electrolyte-free water(P < .001) and volume offluid (P < .001) in casescompared with controls.

Armon et al 200831 Cross-sectionalsurvey

7d-16 y, 99, hospitalizedchildren receiving IV fluids

Hypotonic vs isotonic fluids(multiple types)

Fifteen of 21 children withNa <135 mmol/L receivedhypotonic fluids; 20% ofchildren had no electrolytesin 48 h.

Au et al 200832 Retrospective cohort 1 mo-23 y, 145, postoperative,PICU

Hypotonic vs isotonic fluids(multiple types)

Na <130 mmol/L in 15 of 116in the hypotonic group and 1of 29 in the isotonic group.

Bar�on et al 200933 Prospective cohort 1-60 mo, 35, lower respiratorytract infection

0.225% saline at maintenance;no comparison

Eleven of 35 had a $4 mEq dropin serum Na.

Eulmesekian et al 201034 Prospective cohort Mean 9.8 y, 81, surgical PICU Na 40 mmol + KCl 20 mmol Na <136 mmol/L in 17 of 81 (21%)at 12 h and in 15 of 48 (31%) at24 h (P < .001).

Pemde et al 201026 Randomized controlledtrial

3 mo-5 y, 92, CNS infection Group A (D5NS) vs group B(D5 0.45% saline) vs group C(D5 0.18% saline)

Na<135 mmol/L in 7.1% of group A,46.1% of group B, and 60.7% ofgroup C.

Hanna and Saberi 201035 Retrospective cohort 1 mo-12 y, 124, gastroenteritis Hypotonic vs isotonic fluids(multiple types)

Hyponatremia in 18.5%; no associationbetween hyponatremia and IVfluid type.

Cladis et al 201136 Retrospective cohort <18 y, 72, all postoperativecranial vault remodeling

D5 0.5% or 0.2% salinevs 0.9% saline or LR

Hyponatremia (Na #134 mmol/L) in30.6% (22 of 72).

Kanda et al 201137 Prospective cohort 2-21 y, 60, postrenal biopsy 0.6% saline vs 0.9% saline Hyponatremia (<137 mmol/L) in 60%(6 of 10) of those receivinghypotonic fluids with increasedAVP vs 0% of those receivingisotonic fluids with increased AVP.

AVP, arginine vasopressin; D, dextrose; LR, lactated Ringer’s solution; Na, sodium; NS, normal saline.

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. -, No. -

7.e1 Foster, Tom, and Hill

Figure 1. Schematic of search protocol and results of systematic review. DKA, diabetic ketoacidosis; DI, diabetes insipidus.

- 2014 ORIGINAL ARTICLES

Hypotonic versus Isotonic Fluids in Hospitalized Children: A Systematic Review and Meta-Analysis 7.e2