Diagnosing Clinically Significant Dehydration in Children with Acute

Gastroenteritis Using Noninvasive Methods: A Meta-Analysis

Stephen B. Freedman, MDCM, MSc1, Ben Vandermeer, MSc2, Andrea Milne, MLIS2, and Lisa Hartling, PhD2, on behalf of the
Pediatric Emergency Research Canada Gastroenteritis Study Group*

Objective To determine the most accurate, noninvasive method of assessing dehydration.

Study design The following data sources were searched: electronic databases, gray literature, scientific meet-
ings, reference lists, and authors of unpublished studies. Eligible studies were comparative outpatient evaluations
that used an accepted reference standard and were conducted in developed countries in children aged <18 years
with gastroenteritis. Data extraction was completed independently by multiple reviewers before a consensus was
made.
Results Nine studies that included 1039 participants were identified. The 4-item Clinical Dehydration Scale (CDS),
the “Gorelick” score, and unstructured physician assessment were evaluated in 3, 2, and 5 studies, respectively.
Bedside ultrasound, capillary digital videography, and urinary measurements were each evaluated in one study.
The CDS had a positive likelihood ratio (LR) range of 1.87-11.79 and a negative LR range of 0.30-0.71 to predict
6% dehydration. When combined with the 4-item Gorelick Score, the positive LR was 1.93 (95% CI 1.07-3.49)
and negative LR was of 0.40 (95% CI 0.24-0.68). Unstructured dehydration assessment had a pooled positive
LR of 2.13 (95% CI 1.33-3.44) and negative LR of 0.48 (95% CI 0.28-0.82) to detect $5% dehydration.
Conclusions Overall, the clinical scales evaluated provide some improved diagnostic accuracy. However, test
characteristics indicate that their ability to identify children both with and without dehydration is suboptimal. Current
evidence does not support the routine use of ultrasound or urinalysis to determine dehydration severity. (J Pediatr
2015;-:---).

T
he cornerstone of gastroenteritis management is the assessment of dehydration, with therapy instituted based on
severity.1,2 However, dehydration is difficult to determine clinically,3 and change in body weight remains the “gold stan-
dard.”3 Unfortunately, recent well weights are rarely available,4 and the inaccuracy of available tests limits the ability of
clinicians to estimate the exact degree of dehydration.3 Consequently, research has focused on noninvasive methods of assessing
dehydration (eg, clinical scores,5-8 bedside ultrasound,4,9,10 urine ketones11). Scores, by using combinations of examination
findings, may perform better than individual signs at predicting dehydration.3 Popular examples include the “Gorelick”12
and Clinical Dehydration Scales (CDSs).5 These scales have been adopted, yet their ability to predict severe dehydration is sub-
optimal.13 For example, in a recent report, the 4- and 10-point Gorelick scale had sensitivities of only 64% and 21%, respec-
tively, for severe dehydration.13 Similarly, the ability of bedside ultrasound to assess intravascular volume status remains a topic
of debate.14 Conflicting opinions may relate to the study population and outcome measures used.
A systematic review and meta-analysis focused on research in developed countries can enhance the integration of evidence
into clinical care in such countries. A key focus of quality improvement efforts in children with gastroenteritis in developed
countries is to promote the use of oral rehydration therapy when appropriate in order to minimize the unnecessary adminis-
tration of intravenous fluids.15 Because diagnostic test characteristics (eg, predictive values) are dependent on the prevalence of
disease, evaluating tests of dehydration in the context of developed countries is important. With guidelines recommending that
therapy be tailored to clinical scores16,17 and bedside ultrasound becoming a
standard technology in pediatric emergency departments (EDs) across North
America,18 their roles must be defined.19,20 Thus, we conducted a systematic re- 1
From the Sections of Pediatric Emergency Medicine
and Gastroenterology, Alberta Children’s Hospital,
view of studies in which investigators evaluated the diagnostic test accuracy of Alberta Children’s Hospital Research Institute, University
2
noninvasive methods of dehydration assessment in developed countries. of Calgary, Calgary, Alberta, Canada; and Alberta
Research Centre for Health Evidence, Department of
Pediatrics, University of Alberta, Edmonton, Alberta,
Canada
*List of members of the Pediatric Emergency Research
Canada Gastroenteritis Study Group is available at www.
jpeds.com (Appendix).
Supported by the Canadian Institutes of Health Research
(Knowledge Synthesis grant 262955). L.H. holds a New
Ao Aorta Investigator Salary Award from the Canadian Institutes of
CDS Clinical Dehydration Scale Health Research. The authors declare no conflicts of in-
terest.
ED Emergency department
IVC Inferior vena cava 0022-3476/$ - see front matter. Copyright ª 2015 Elsevier Inc.
LR Likelihood ratio All rights reserved.
http://dx.doi.org/10.1016/j.jpeds.2014.12.029

1
THE JOURNAL OF PEDIATRICS  www.jpeds.com Vol. -, No. -

comes (eg diagnostic accuracy), source of funding, and re-

Methods
sults. Extracted data were entered into Microsoft Excel
(Microsoft, Redmond, Washington) worksheets. Disagree-
We followed a standard protocol for the conduct and report-
ments were resolved by consensus, or involving a third
ing of systematic reviews and meta-analyses, which was in
reviewer as required. The Quality Assessment of Diagnostic
keeping with the Preferred Reporting Items for Systematic re-
Accuracy Studies 2 tool27 was used to assess the methodo-
views and Meta-Analyses guidelines.21,22 An experienced
logic quality of the relevant studies. The Quality Assessment
medical librarian developed a search strategy in collaboration
of Diagnostic Accuracy Studies 2 includes 4 domains: patient
with the research team to identify studies examining the diag-
selection, index test, reference standard, and flow of patients
nostic accuracy of noninvasive methods of assessing dehydra-
through the study and the timing of the index tests and refer-
tion. We (1) systematically searched MEDLINE (1946 to
ence standard (flow and timing).28 Assessments regarding
April 2013), EMBASE (1980 to April 2013), the Cochrane
bias and applicability are made for each domain. Bias is as-
Central Register of Controlled Trials (April 2013) via the
sessed as low, unclear, or high risk; applicability is assessed
OvidSP platform, PubMed via the National Library of Med-
as low, unclear, or high concerns. Quality assessment was
icine (last 180 days), and for gray literature; (2) hand-
completed independently by 2 reviewers. Disagreements
searched appropriate journals and major, relevant scientific
were resolved by consensus, or involving a third reviewer as
meetings (Society for Pediatric Research 2010-2012, Amer-
required.
ican Academy of Pediatrics 2010-2012, Canadian Pediatric
We developed evidence tables to describe the studies
Society 2010-2012, and International Conference on Emer-
including information on design features, methodologic
gency Medicine 2012); (3) checked reference lists of relevant
quality, study populations, sample size, settings, dehydration
studies; and (4) contacted primary authors of published and
scores, and comparisons. For each of the included studies we
unpublished studies. The MEDLINE search strategy is ap-
extracted the raw data regarding true and false positives and
pended (Table I; available at www.jpeds.com). The search
negatives and constructed 2  2 tables to calculate sensitivity,
was not restricted by language or publication status. We
specificity, and likelihood ratios (LRs). Sensitivity and spec-
ran an updated search of the electronic databases in
ificity are measures of test accuracy. LRs are used to estimate
October 2014 to identify studies published after the first
the increased or decreased probability of disease (ie, dehydra-
search; no eligible studies were identified. All studies
tion) for a patient and can be used to refine clinical judgment.
contained in previous relevant systematic reviews were
The larger the positive LR, the greater the accuracy of the test
screened for inclusion.
and the greater the likelihood of disease after a positive test
Search result titles and abstracts were screened indepen-
result. In contrast, the smaller the negative LR, the lower
dently by 2 reviewers to identify potentially relevant citations.
the likelihood of disease after a negative test result.29 Sensitiv-
They were excluded when the title/abstract did not identify
ities, specificities, LRs, and predictive values are presented in
that the article addressed the accuracy of a noninvasive
a summary table that includes all dehydration assessment
method of assessing dehydration. The full text of all poten-
methods. We planned to analyze data using hierarchical sum-
tially relevant citations was obtained and assessed for inclu-
mary receiver-operating curves; however, an insufficient
sion by 2 independent reviewers using standard, predefined
number of studies examining any given test were identified
eligibility criteria. Disagreements were resolved by consensus.
to enable the use of this approach.30 Consequently, we
Decisions regarding inclusion and reasons for exclusion were
plotted the sensitivity and specificity of the individual studies
documented. Original studies were included if they: (1) eval-
in a receiver-operating curve space to graphically display the
uated children <18 years of age suspected to have dehydra-
relative accuracy of the different measures. We pooled LRs
tion; (2) examined the diagnostic accuracy of a noninvasive
using Maentel-Hansel methods and random effects models.
method of dehydration assessment against percent change
We were unable to formally assess for publication bias
in body weight between acute presentation and stable, rehy-
because of the small numbers of studies examining any given
drated, well weight (Table II)12; (3) were conducted in a
test.
developed country as defined by the United Nations in
2011—Australia, Canada, Europe, Japan, New Zealand, and
the US23; and (4) were conducted in an ED or similar Results
clinical setting. We included studies in which authors
focused on children with acute gastroenteritis. Comparative The electronic database search identified 1454 citations;
studies meeting the aforementioned criteria were included. 66 were considered potentially relevant based on their title/
Review articles were excluded. abstract (Figure 1; available at www.jpeds.com). Of these,
As is commonly performed, 1 reviewer extracted data us- 4 met inclusion. Five additional studies meeting eligibility
ing a structured form. Verification was performed by a sec- criteria were identified by reviewing the references of
ond reviewer for accuracy and completeness.24-26 The relevant studies (Table II). The median year of publication
following items were extracted: study characteristics (eg, was 2007. The 9 eligible studies included 1293 participants,
date of publication, clinical setting, country), participants of whom 1039 (80%) had both the diagnostic evaluation
(eg, age, sex), dehydration scores and comparisons, out- and the reference gold standard performed. Eight studies

2 Freedman et al
A Meta-Analysis
Diagnosing Clinically Significant Dehydration in Children with Acute Gastroenteritis Using Noninvasive Methods:

- 2015
Table II. Characteristics of studies included in systematic review
Author,
reference, Study design Number enrolled Setting Age, y, mean (SD) Dehydration
year, country (enrollment) (complete data) (patient type) or median [IQR] severity, % (n) Inclusion criteria Index test(s) evaluated Dehydration reference standard
Chen et al,4 prospective 112 (71) Hospital ED 5.3 (4.3)  <5% (43)  6 months to 18 years  Ultrasound (IVC:Ao)  Repeat weight once asymptomatic
2010, US cohort (NR) (outpatient)  $5% (28)  Acute gastroenteritis  Physician assessment for 48 h, 7-10 d postindex visit
(diarrhea with or without  Difference between ill weight and
vomiting, onset within well weight was used to calculate
7 days) % fluid loss at index visit
Gravel et al,7 prospective cohort 264 (219) Hospital ED 1.3 [0.75-2.3]  <3% (150)  1 mo to 5 y  CDS  Repeat weight once asymptomatic
2010, Canada (convenience) (outpatient)  3%-6% (40)  Vomiting or diarrhea for 48 h, 5-7 d postindex visit
 >6% (29) for <7 d  Difference between ill weight and
 Not requiring immediate well weight was used to calculate
physician assessment % fluid loss at index visit
(ie, severe dehydration
excluded)
Gorelick et al,12 prospective 225 (186) Hospital ED 1.1 [NR]  <5% (123)  1 month to 5 years  Gorelick scale  Stable weight reached when
1997, US cohort (outpatient)  5%-10% (52)  Vomiting, diarrhea, or 2 consecutive weights differed
(convenience)  $10% (11) poor fluid intake by <2%
 Mean of final 2 weights was
the “well” weight
 Subjects weighed at end of ED visit,
and again 48-72 h later, after
symptoms had resolved
 Daily visits occurred until stable
weight achieved
 Difference between ill weight and
well weight was used to calculate
% fluid loss at index visit
Jauregui et al,35 prospective cohort 148 (113) Hospital ED 6 [range: 1  <5% (101)  #18 y  CDS  Child was weighed before
2014, US (non-consecutive) (outpatient) mo-18 y]  $5% (12)  Acute diarrhea or  Gorelick scale intravenous fluid administration
vomiting, or suspicion  World Health and again at completion of ED visit
of dehydration Organization scale  “Well” weight was recorded upon
 Administered  Physician assessment discharge from ED or in hospitalized
intravenous rehydration children was recorded prior to
discharge from inpatient unit
 Difference between ill weight and
well weight was used to calculate
% fluid loss at index visit
Mackenzie et al,32 prospective 102 (102) Hospital Ward NR  <5% (76)  0-4 y  Physician assessment  Degree of dehydration
1989, Australia cohort (NR) (inpatient)  5%-10% (26)  Gastroenteritis was calculated by subtracting
 Previously well prerehydration weight from the
 Clinically assessed postrehydration weight
to be 5%-10% dehydrated  Children were reweighed daily

ORIGINAL ARTICLES
until discharge
(continued )
3
4

THE JOURNAL OF PEDIATRICS

Table II. Continued
Author,
reference, Study design Number enrolled Setting Age, y, mean (SD) Dehydration
year, country (enrollment) (complete data) (patient type) or median [IQR] severity, % (n) Inclusion criteria Index test(s) evaluated Dehydration reference standard
Parkin et al,31 prospective cohort 141 (93) Hospital ED 1.5 [range:  <3% (49)  1-36 mo  CDS  Children were weighed
2010, Canada (convenience) (outpatient) 0.2-2.9]  $3-<6% (34)  Acute gastroenteritis before initiation of rehydration
 $6% (10) therapy and again when ED or
inpatient physician (if child
admitted to hospital) determined
that rehydration was complete
 Difference between ill weight and
well weight was used to calculate
% fluid loss at index visit
Shavit et al,34 prospective cohort 87 (83) Hospital ED 1.5 [0.9-2.8]  <5% (70)  1 mo-5 y  Digitally measured  Stable weight reached when
2006, Canada (convenience) (outpatient)  $5%-10% (12)  Acute gastroenteritis: capillary-refill time 2 consecutive weights differed
 $10% (1) diarrhea (with or without  Physician assessment by <2%


vomiting) #5 d  Mean of final 2 weights was

www.jpeds.com
 Some dehydration the “well” weight
 Subjects weighed at end of ED visit
and again 48-72 hours later, after
symptoms had resolved
 Daily visits occurred until stable
weight achieved
 Difference between ill weight and
well weight was used to calculate
% fluid loss at index visit
Steiner et al,11 prospective cohort 79 (75) Hospital ED 1.4 (NR)  <3% (25)  3-36 mo  Urine-specific gravity  Degree of dehydration was
2007, US (convenience) (outpatient)  3%-5% (34)  Acute gastroenteritis  Urine ketones calculated by subtracting
 $5% (16)  Suspected moderate prerehydration weight from the
dehydration postrehydration weight
 Administered  Difference between ill weight and
intravenous rehydration well weight was used to calculate
% fluid loss at index visit
Vega et al,33 prospective cohort 135 (97) Hospital ED 3.2 (range:  #5% (50)  All children (age not  Physician assessment  Subjects weighed within 72 hours
1997, US (consecutive) (outpatient) 2 wk-15 y)  6%-10% (30) specified) of index visit
 >10% (17)  Administered  If acute illness had resolved that
intravenous rehydration weight was deemed the “well”
 Gastroenteritis (85%); weight
stomatitis (4%);  If still ill, ongoing follow-up was
other (11%) scheduled
 Difference between ill weight and
well weight was used to calculate
% fluid loss at index visit

NR, not reported.

Sample size numbers reflect the total number of participants completing follow-up enabling a gold standard assessment of dehydration.

Vol. -, No. -
Freedman et al
- 2015
recruited patients presenting for ED care. Six studies
classified dehydration employing a <5%; 5%-10%, and
>10% scale; 3 used <3%; 3%-6%, and >6% as their
reference standard to reflect mild/moderate/severe
dehydration. Studies displayed significant heterogeneity in
terms of participant age, both at the younger and older
extremes.
Quality of Studies
Patient selection was rated as low risk of bias in 1 study; the
remaining studies were unclear or at high risk of bias
(Table III and Figure 2; available at www.jpeds.com). Six
studies had unclear or high concerns about applicability
largely because information on patient selection was not
reported. The index test domain was at low risk of bias
in 5 studies. One study had high risk of bias because
researchers did not use prespecified thresholds for
classifying dehydration severity. Three studies were
assessed as unclear because they did not provide sufficient
information to determine whether the researchers
prespecified thresholds or whether the reference standard
results were known when the index test was interpreted.
All studies had low concerns regarding applicability. The
domain of the reference standard was at low risk of bias
for 5 studies; 4 were rated as unclear because it was
unknown whether the reference standard results were
interpreted without knowledge of the index test result or
it was unclear when the well-weights were assessed. All
studies had low concerns regarding applicability for this
domain. Flow and timing was at low risk of bias for 4
studies; 5 were high risk of bias because a significant
proportion of participants were lost to follow-up or had
incomplete measurements.
Noninvasive Assessments
The CDS was evaluated in 3 studies. This 4-item scale was
derived in a cohort of 137 children 5(Tables IV and V;
Table V; available at www.jpeds.com). The authors used
formal approaches to derive the scale. Although diagnostic
accuracy was not reported in the derivation sample,
subsequent work has conducted such analyses assigning
scores of 0, 1, or 2 for each item that is ranked along a
continuum from normal to significantly abnormal.7,31
These studies (Table IV) revealed that a score of 0 had a
positive LR of 1.64-2.19 and negative LR of 0.79-0.84 to
predict <3% dehydration. The ability of the score to
predict $5% (or >6%) dehydration had positive LRs
ranging between 1.87 and 11.79 and negative LR ranged
between 0.30 and 0.71 (Figure 3).
The “Gorelick” scale, which formalized the dehydration
evaluation via clinical features, uses 4- and 10-item scales.12
Because of the overlap in clinical features between the CDS
and the 4-item Gorelick score, these scores were combined
to generate pooled estimates with positive LR = 1.93 (95%
CI 1.07-3.49) and negative LR = 0.40 (95% CI 0.24-0.68).
The overall accuracy of these scales ranged between 57%
and 88% (Table IV).
Diagnosing Clinically Significant Dehydration in Children with Acute Gastroenteritis Using Noninvasive Methods:
A Meta-Analysis
ORIGINAL ARTICLES
Physician clinical assessment of dehydration has been eval-
uated by the use of several approaches. These include an
overall hydration assessment by: (1) junior doctors, who
were instructed to classify children as being 5%-10% dehy-
drated based on their hospital’s guideline32; (2) the fellow
or attending physician, by using a 9-item “standard clinical
scale”33; (3) the attending physician, blinded to ultrasound
findings and using an ordinal scale (<5%, 5%-10%,
>10%)4; (4) the pediatric emergency physician, who esti-
mated the child’s degree of dehydration by using a 7-point
Likert scale (very mild to very severe)34; and (5) having the
treating pediatric emergency medicine attending physicians
record their clinical impression (1-10 scale) for percent dehy-
dration.35
Excluding a study with 100% sensitivity because it
included only children deemed to be 5%-10% dehydrated
(ie, excluded children who might have been false negatives),32
the sensitivity of this approach to detect $5% dehydration
ranged from 33% to 78%. The pooled results from 4 studies
for positive and negative LRs were 2.13 (95% CI 1.33-3.44)
and 0.48 (95% CI 0.28-0.82), respectively. Overall accuracy
ranged between 25% and 81% (Table IV).
Chen et al4 evaluated the accuracy of an inferior vena cava
(IVC) to aorta (Ao) ratio of <0.8 to predict $5% weight loss.
Although ultrasound had a negative LR value of 0.27, the
positive LR of 1.95 was significantly lower than both the
CDS and the Gorelick scales for identifying >6% dehydra-
tion. A single study also evaluated the ability of digital
capillary refill to predict $5% body weight change. The au-
thors reported an positive LR of 11.67 and negative LR of
0.00 and the greatest overall accuracy of any measure evalu-
ated (93%).34 The sensitivities and specificities of all assess-
ment methods identified are depicted in an ROC plot
(Figure 3).
Interobserver Assessments (ie, Reproducibility)
These were performed in 4 of the 9 studies and varied in size
related to the entire cohort. Kappa values were lowest for the
clinical measures—0.65 for the CDS and 0.69 for the Gorelick
scale.7,12 The values were greater for ultrasound (Pearson
correlation coefficient = 0.76)4 and digital capillary refill (in-
traclass correlation = 0.99).34
Discussion
The practice of clinical medicine is founded on the reliability
of history and physical examination to guide management.
However, the evidence regarding the diagnostic accuracy of
clinical examination is often lacking; pediatric dehydration is
no exception. The use of the CDS or the Gorelick scale both
consistently reported accuracy values of >80% in identifying
significant dehydration, although lower values were noted by
Jauregui et al.35 Overall, the clinical scales evaluated provide
some improved diagnostic accuracy. However, test character-
istics indicate that a substantial gap remains in how well the
scores predict the outcome of interest. Similarly, bedside ultra-
sound, which may aid in ruling out dehydration (LR negative
5
6

THE JOURNAL OF PEDIATRICS

Table IV. Results of studies included in systematic review (reference standard is weight change)
Predictive values LRs
Author, year, country N Index test Sensitivity, % (95% CI) Specificity, % (95% CI) Positive Negative Positive Negative Accuracy
CDS (4 items, 8 points)
Parkin et al,31 2010, Canada 93 CDS score = 0 0.32 (0.21-0.46) 0.85 (0.69-0.95) 79.17 42.03 2.19 0.79 51.61
<3% dehydration
93 CDS score = 1-4 0.67 (0.45-0.84) 0.42 (0.30-0.55) 28.57 78.38 1.15 0.79 48.39
3%-6% dehydration
93 CDS score = 5-8 0.50 (0.19-0.81) 0.90 (0.82-0.96) 38.46 93.75 5.19 0.55 86.02
>6% dehydration
Gravel et al,7 2010, Canada 219 CDS score = 0 0.33 (0.26-0.41) 0.80 (0.68-0.88) 78.13 35.48 1.64 0.84 47.95
<3% dehydration
219 CDS score = 1-4 0.75 (0.59-0.87) 0.38 (0.31-0.46) 21.28 87.18 1.21 0.66 44.75
3–6% dehydration
219 CDS score = 5-8 0.31 (0.15-0.51) 0.97 (0.94-0.99) 64.29 90.24 11.79 0.71 88.58


>6% dehydration
35
CDS score $2

www.jpeds.com
Jauregui et al, 2014, US 113 0.83 (0.52-0.98) 0.55 (0.45-0.65) 18.18 96.55 1.87 0.30 58.41
$5% dehydration
Physician clinical assessment
Vega et al,33 1997, US 97 Mild 0.74 (0.60-0.85) 0.70 (0.55-0.83) 72.55 71.74 2.48 0.37 72.16
<5% dehydration
97 Moderate 0.33 (0.17-0.53) 0.84 (0.73-0.92) 47.62 73.68 2.03 0.80 68.04
6%–10% dehydration
97 Severe 0.71 (0.44-0.90) 0.84 (0.74-0.91) 48.00 93.06 4.34 0.35 81.44
$10% dehydration
Chen et al,4 2010, US 71 Significant 0.78 (0.61-0.90) 0.51 (0.34-0.69) 62.22 69.23 1.60 0.43 64.79
$5% dehydration
Jauregui et al,37 2014, US 113 $5% dehydration 0.42 (0.15-0.72) 0.65 (0.55-0.75) 12.50 90.41 1.20 0.89 62.83
Mackenzie et al,37 1989, Australia 102 Moderate 5%-10% dehydration 1.00 (0.87-1.00) 0.00 (0.00-0.05) 25.49 n/a 1.00 n/a 25.49
Shavit et al,34 2006, Canada 83 Overall clinical assessment on 7-point Likert scale 0.77 (0.46-0.95) 0.81 (0.70-0.90) 43.48 95.00 4.14 0.28 80.72
Gorelick score
Gorelick et al,12 1997, US 186 4-item model score $2 0.79 (0.67-0.89) 0.87 (0.80-0.93) 75.76 89.43 6.25 0.24 84.66
$5% dehydration
186 4-item model score $3 0.82 (0.48-0.98) 0.83 (0.77-0.88) 23.08 98.67 4.85 0.22 83.07
$ 10% dehydration
186 10-item model score $3 0.87 (0.77-0.94) 0.82 (0.74-0.88) 71.43 92.66 4.88 0.15 83.87
$5% dehydration
186 10-item model score $7 0.82 (0.48-0.98) 0.87 (0.81-0.91) 28.13 98.70 6.23 0.21 86.56
$10% dehydration
Jauregui et al,35 2014, US 113 10-item score $2 0.75 (0.43-0.95) 0.55 (0.45-0.65) 16.67 94.92 1.68 0.45 57.52
$5% dehydration
World Health Organization Scale
Jauregui et al,35 2014, US 113 4-item scale $2 items 0.25 (0.05-0.57) 0.84 (0.76-0.91) 15.79 90.43 1.58 0.89 77.88
$5% dehydration
Bedside ultrasound
Chen et al,4 2010, US 71 IVC:Ao <0.80 on bedside ultrasound 0.85 (0.68-0.95) 0.56 (0.40-0.72) 62.22 81.48 1.95 0.27 69.44
Digital capillary refill time

Vol. -, No. -
Freedman et al

Shavit et al,34 2006, Canada 83 DCRT $0.4 s 1.00 (0.75-1.00) 0.91 (0.82-0.97) 68.42 100.00 11.67 0.00 92.77
Laboratory measures
Steiner et al,11 2007, US 75 Urine-specific gravity, 1.030 (3%) 0.64 (0.49-0.77) 0.40 (0.21-0.61) 68.09 35.71 1.07 0.90 56.00
Steiner et al,11 2007, US 75 Urine-gravity, 1.030 (5%) 0.38 (0.15-0.65) 0.31 (0.19-0.44) 12.77 64.29 0.54 2.05 32.00

DCRT, digital capillary refill time; n/a, (not applicable) due to 0 cell counts (ie, no false negatives or true negatives).
- 2015
Figure 3. Receiver-operating curve plot showing combined sensitivity and specificity of the different diagnostic tests. WHO,
World Health Organization.
of 0.27), should not be used to confirm the presence of dehy-
dration (LR positive of 1.95). The most promising interven-
tion we identified was the use of digital videography
capillary refill time; given the paucity of data, however, further
research is required to confirm the preliminary findings.
Our findings are congruent with a previous review that
included studies in which authors evaluated dehydration
assessment in developed and developing countries.3 This re-
view concluded that the quantity and quality of research on
dehydration assessment is extremely limited, with most
studies being of only moderate sample size and at significant
risk of patient selection bias. Many were conducted at pedi-
atric institutions and enrolled exclusively children deemed
to be dehydrated and those needing rehydration. This high-
lights how study populations may not reflect the real-world
situation, where clinicians must identify the small number
of children with significant dehydration from those who
are not dehydrated.
When considering our findings, it is important to
consider what the most important measure of test perfor-
Diagnosing Clinically Significant Dehydration in Children with Acute Gastroenteritis Using Noninvasive Methods:
A Meta-Analysis
ORIGINAL ARTICLES
mance is—overall test accuracy (true positive + true nega-
tive) vs sensitivity or negative LR. The decision reached is
likely based on one’s perspective on the number to over-
diagnose (ie, unnecessary intravenous fluids) to identify
one case of severe dehydration. Clinicians must consider
the limited importance of distinguishing small increments
of dehydration. This issue is highlighted in guidelines pub-
lished by the Centers for Disease Control, which group
children into none/minimal (<3%-5%), mild to moderate
(3%-9%), and severe ($9%) dehydration and tie manage-
ment to these categories.2
Ultrasound has been touted as having the potential to serve
as a useful adjunct for this clinical dilemma.36 In the single
study identified,4 a pair of investigators obtained transverse
images of the IVC and the Ao and then calculated their ratio
(IVC/Ao). The authors reported that an IVC/Ao cut-off of 0.8
produced a sensitivity of 86%, a specificity of 56%, and a pos-
itive predictive value of 62% for the diagnosis of significant
dehydration. These characteristics are not superior to clinical
dehydration scales, and less than two-thirds of test-positive
7
THE JOURNAL OF PEDIATRICS  www.jpeds.com
patients will have significant dehydration. Thus, there is
insufficient evidence to support the routine use of ultrasound
due to the potential to result in “overtreatment” if test char-
acteristics are not considered. More recent work has also
concluded that the IVC and Ao measurements by bedside ul-
trasound are not reliable indicators of intravascular volume as
determined by central venous pressure.14 Digital capillary
refill time has only been described in a single study.34
The present study has several potential limitations. The
greatest challenge in quantifying dehydration severity is the
gold standard. Although this study required, in keeping with
expert opinion and previous studies,3,5,12,32,33 that postillness
weight be used as the gold standard, the included studies
differed significantly in the precise day used as the reference
measure of stable post-illness weight (Table II).4,12,33,37 An
additional concern relates to the timing within a given day
of rehydration weights, which are influenced by stooling,
voiding, and feeding, especially in infants.3 Nonetheless, this
study followed the use of this gold standard as the criteria
for defining true percent dehydration. Although this study
followed accepted methodologic standards for the conduct
of systematic reviews of diagnostic test accuracy including a
comprehensive search of the literature, there remains the
possibility that some unpublished studies may not have been
identified. Although formally evaluating publication bias
through the use of funnel plots and statistical tests would be
ideal, due to insufficient numbers of studies included in this
review, a formal analysis of publication bias was not possible.
Overall, the accuracy of clinical assessments of dehydra-
tion is suboptimal. Current evidence does not support the
routine use of ultrasound or urinalysis to determine dehydra-
tion severity. Technological innovations, such as digital
capillary refill, require further evaluation. The success of
such work could enable clinicians to improve the care pro-
vided to children at risk for dehydration. n
We thank the following for their contributions to this project: Melanie
Muise and Dion Pasichnyk for their assistance with screening, study se-
lection, data extraction, and verification, and Donna Dryden, MD, for
assistance with quality assessment.
Submitted for publication Aug 29, 2014; last revision received Oct 31, 2014;
accepted Dec 12, 2014.
References
1. Guarino A, Albano F, Ashkenazi S, Gendrel D, Hoekstra JH,
Shamir R, et al. European Society for Paediatric Gastroenterology,
Hepatology, and Nutrition/European Society for Paediatric Infectious
Diseases evidence-based guidelines for the management of acute
gastroenteritis in children in Europe. J Pediatr Gastroenterol Nutr
2008;46(Suppl 2):S81-122.
2. King CK, Glass R, Bresee JS, Duggan C. Managing acute gastroenteritis
among children: oral rehydration, maintenance, and nutritional therapy.
MMWR Recomm Rep 2003;52:1-16.
3. Steiner MJ, DeWalt DA, Byerley JS. Is this child dehydrated? JAMA 2004;
291:2746-54.
4. Chen L, Hsiao A, Langhan M, Riera A, Santucci KA. Use of bedside ul-
trasound to assess degree of dehydration in children with gastroenteritis.
Acad Emerg Med 2010;17:1042-7.
8 Freedman et al
Vol. -, No. -
5. Friedman JN, Goldman RD, Srivastava R, Parkin PC. Development of a
clinical dehydration scale for use in children between 1 and 36 months of
age. J Pediatr 2004;145:201-7.
6. Bailey B, Gravel J, Goldman RD, Friedman JN, Parkin PC. External vali-
dation of the clinical dehydration scale for children with acute gastroen-
teritis. Acad Emerg Med 2010;17:583-8.
7. Gravel J, Manzano S, Guimont C, Lacroix L, Gervaix A, Bailey B. Multi-
center validation of the clinical dehydration scale for children [in
French]. Arch Pediatr 2010;17:1645-51.
8. Goldman RD, Friedman JN, Parkin PC. Validation of the clinical dehy-
dration scale for children with acute gastroenteritis. Pediatrics 2008;122:
545-9.
9. Levine AC, Shah SP, Umulisa I, Munyaneza RB, Dushimiyimana JM,
Stegmann K, et al. Ultrasound assessment of severe dehydration in chil-
dren with diarrhea and vomiting. Acad Emerg Med 2010;17:1035-41.
10. Kosiak W, Swieton D, Piskunowicz M. Sonographic inferior vena cava/
aorta diameter index, a new approach to the body fluid status assessment
in children and young adults in emergency ultrasound—preliminary
study. Am J Emerg Med 2008;26:320-5.
11. Steiner MJ, Nager AL, Wang VJ. Urine specific gravity and other urinary
indices: inaccurate tests for dehydration. Pediatr Emerg Care 2007;23:
298-303.
12. Gorelick MH, Shaw KN, Murphy KO. Validity and reliability of clinical
signs in the diagnosis of dehydration in children. Pediatrics 1997;99:E6.
13. Pringle K, Shah SP, Umulisa I, Mark Munyaneza RB,
Dushimiyimana JM, Stegmann K, et al. Comparing the accuracy of the
three popular clinical dehydration scales in children with diarrhea. Int
J Emerg Med 2011;4:58.
14. Ng L, Khine H, Taragin BH, Avner JR, Ushay M, Nunez D. Does bedside
sonographic measurement of the inferior vena cava diameter correlate
with central venous pressure in the assessment of intravascular volume
in children? Pediatr Emerg Care 2013;29:337-41.
15. Santosham M, Keenan EM, Tulloch J, Broun D, Glass R. Oral rehydra-
tion therapy for diarrhea: an example of reverse transfer of technology.
Pediatrics 1997;100:E10.
16. Cincinnatti Children’s Hospital Medical Center Acute Gastroenteritis
Team 2011. Evidence-Based Care Guideline for Prevention and Manage-
ment of Acute Gastroenteritis (AGE) in children age 2 mo to 18 yrs,
http://www.cincinnatichildrens.org/WorkArea/linkit.aspx?LinkIdentifier=
id&ItemID=93672&libID=93365. Accessed October 16, 2014.
17. van den Berg J, Berger MY. Guidelines on acute gastroenteritis in chil-
dren: a critical appraisal of their quality and applicability in primary
care. BMC Fam Pract 2011;12:134.
18. Vieira RL, Bachur R. Bedside ultrasound in pediatric practice. Pediatrics
2014;133:1-3.
19. Levy JA, Bachur RG. Bedside ultrasound in the pediatric emergency
department. Curr Opin Pediatr 2008;20:235-42.
20. Leeson K, Leeson B. Pediatric ultrasound: applications in the emergency
department. Emerg Med Clin North Am 2013;31:809-29.
21. Higgins JPT, Green S. Cochrane Handbook for Systematic Reviews of In-
terventions. Version 5.0.2 [updated September 2009]. The Cochrane
Collaboration, 2009, www.cochrane-handbook.org. Accessed December
23, 2014.
22. Moher D, Liberati A, Tetzlaff J, Altman DG. The Prisma Group.
Preferred reporting items for systematic reviews and meta-analyses:
the PRISMA statement. BMJ 2009;339:b2535.
23. Composition of macro geographical (continental) regions, geographical
sub-regions, and selected economic and other groupings (footnote c).
United Nations Statistics Division revised February 17, 2011 [cited
2011 March 4]. Accessed December 23, 2014, http://unstats.un.org/
unsd/methods/m49/m49regin.htm#ftnc.
24. Seida JC, LeBlanc C, Schouten JR, Mousavi SS, Hartling L,
Vandermeer B, et al. Systematic review: nonoperative and operative
treatments for rotator cuff tears. Ann Intern Med 2010;153:246-55.
25. Hartling L, Dryden DM, Guthrie A, Muise M, Vandermeer B,
Donovan L. Benefits and harms of treating gestational diabetes
mellitus: a systematic review and meta-analysis for the U.S. Preven-
tive Services Task Force and the National Institutes of Health
- 2015
Office of Medical Applications of Research. Ann Intern Med 2013;
159:123-9.
26. Hartling L, Fernandes RM, Bialy L, Milne A, Johnson D, Plint A,
et al. Steroids and bronchodilators for acute bronchiolitis in the first
two years of life: systematic review and meta-analysis. BMJ 2011;342:
d1714.
27. Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB,
et al. QUADAS-2: a revised tool for the quality assessment of diagnostic
accuracy studies. Ann Intern Med 2011;155:529-36.
28. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The devel-
opment of QUADAS: a tool for the quality assessment of studies of diag-
nostic accuracy included in systematic reviews. BMC Med Res Methodol
2003;3:25.
29. Jaeschke R, Guyatt GH, Sackett DL. Users’ guides to the medical litera-
ture. III. How to use an article about a diagnostic test. B. What are the
results and will they help me in caring for my patients? The Evidence-
Based Medicine Working Group. JAMA 1994;271:703-7.
30. Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM,
Zwinderman AH. Bivariate analysis of sensitivity and specificity pro-
duces informative summary measures in diagnostic reviews. J Clin Epi-
demiol 2005;58:982-90.
Diagnosing Clinically Significant Dehydration in Children with Acute Gastroenteritis Using Noninvasive Methods:
A Meta-Analysis
ORIGINAL ARTICLES
31. Parkin PC, Macarthur C, Khambalia A, Goldman RD, Friedman JN.
Clinical and laboratory assessment of dehydration severity in children
with acute gastroenteritis. Clin Pediatr (Phila) 2010;49:235-9.
32. Mackenzie A, Barnes G, Shann F. Clinical signs of dehydration in chil-
dren. Lancet 1989;2:605-7.
33. Vega RM, Avner JR. A prospective study of the usefulness of clinical and
laboratory parameters for predicting percentage of dehydration in chil-
dren. Pediatr Emerg Care 1997;13:179-82.
34. Shavit I, Brant R, Nijssen-Jordan C, Galbraith R, Johnson DW. A novel
imaging technique to measure capillary-refill time: improving diagnostic
accuracy for dehydration in young children with gastroenteritis. Pediat-
rics 2006;118:2402-8.
35. Jauregui J, Nelson D, Choo E, Stearns B, Levine AC, Liebmann O, et al.
External validation and comparison of three pediatric clinical dehydra-
tion scales. PLoS One 2014;9:e95739.
36. Chen L, Kim Y, Santucci KA. Use of ultrasound measurement of the infe-
rior vena cava diameter as an objective tool in the assessment of children
with clinical dehydration. Acad Emerg Med 2007;14:841-5.
37. Mackenzie A, Barnes G. Randomised controlled trial comparing oral and
intravenous rehydration therapy in children with diarrhoea. BMJ 1991;
303:393-6.
9
THE JOURNAL OF PEDIATRICS  www.jpeds.com Vol. -, No. -

Appendix

Additional members of the Pediatric Emergency Research

Canada Gastroenteritis Study Group include:
David Johnson, MD (Alberta Children’s Hospital, Univer-
sity of Calgary, Alberta Children’s Hospital Research Insti-
tute, Calgary, AB), Karen Black, MD (University of British
Columbia, Vancouver, BC), Robert Porter, MD (Janeway
Children’s Hospital and Rehabilitation Centre, Memorial
University, St. John’s NL), Gary Joubert, MD (Schulich
School of Medicine & Dentistry, Children’s Health Research
Institute, London Health Sciences Centre, Western Univer-
sity, London, ON), Serge Gouin, MDCM (Universite de
Montreal, Centre Hospitalier Universitaire Sainte-Justine,
Montreal, QC), Quynh Doan, MDCM, MHSc, PhD (Child
& Family Research Institute, BC Children’s Hospital, Univer-
sity of British Columbia, Vancouver, BC), Janie Williamson,
RN (Alberta Children’s Hospital, University of Calgary, Cal-
gary, AB), Lynell Aucoin, RN (Women and Children’s Health
Research Institute, University of Alberta, Edmonton, AB), El-
eanor Fitzpatrick, BScN, MN (IWK Health Centre, Dalhousie
University, Halifax, NS), Mona Jabbour, MD (Children’s
Hospital of Eastern Ontario, Children’s Hospital of Eastern
Ontario Research Institute, University of Ottawa, Ottawa,
ON), and Terry Klassen, MD (Manitoba Institute of Child
Health, Winnipeg Regional Health Authority, University of
Manitoba, Winnipeg, MB).

9.e1 Freedman et al
- 2015 ORIGINAL ARTICLES

Table I. Search strategy for Medline via Ovid

MeSH and keywords for gastroenteritis:
1. exp Gastroenteritis/
2. Rotavirus/
3. Rotavirus Infections/
4. exp Norovirus/
5. exp Diarrhea/
6. Vomiting/
7. gastroenterit*.tw.
8. (diarrhea or diarrhoea).tw.
9. vomit*.tw.
10. gastritis.tw.
11. or/1-10 [MeSH and keywords for Gastro—without dehydration] (267,727)
MeSH and keywords for body fluids, body weight, and dehydration:
12. exp Body Fluids/
13. exp Body Weight/
14. Dehydration/
15. ((change* or loss*) adj3 body weight).tw.
16. ((fluid* or water) adj3 (status or volume)).tw.
17. dehydrat*.tw.
18. or/12-17 [MeSH and keywords for body fluids/weight] (628,497)
MeSH and keywords for diagnostic methods:
19. Point-of-Care Systems/
20. Health status indicators/
21. “Severity of Illness Index”/
22. Diagnostic Techniques, Digestive System/
23. exp Ultrasonography/
24. ((“point of care” or bedside or emergenc*) adj3 (ultrasound or sonograph* or ultrasonograph* or US)).tw.
25. (severit* adj3 (scale* or score* or rating or rate* or test)).tw.
26. or/19-25 [MeSH or keywords r/t US or dehydration scores] (415,155)
27. and/11,18,26 [GE + body weight/fluid + assessment tools] (381)
28. limit 27 to “all child (0 to 18 years)” (199)
29. exp Child/
30. exp Infant/
31. Adolescent/
32. Minors/
33. exp pediatrics/
34. (infant* or child* or adolescen* or teen* or youth* or young or p?ediatric* or juvenile*).tw.
35. or/29-34 [MeSH and keywords for child age group] (3,112,375)
36. and/27,35 [application of child age limit] (209)
37. or/28,36 [combination of Medline + manual age limit filters] (209)
38. remove duplicates from 37 (206)

MeSH, Medical Subject Headings.

Diagnosing Clinically Significant Dehydration in Children with Acute Gastroenteritis Using Noninvasive Methods: 9.e2
A Meta-Analysis
9.e3

THE JOURNAL OF PEDIATRICS

Table III. Methodologic quality of included studies using QUADAS-2
1. Patient selection* 2. Index test* 3. Reference standard* 4. Flow and timing*
b. index
c. risk d. concerns c. risk d. concerns a. likely results c. risk d. concerns c. all included d. concerns
Author, a. b. of of a. reference results of of to not of of in of
year sample exclusion bias applicability not known b. threshold bias applicability classify known bias applicability a. interval b. same standard analysis applicability
Chen No Yes High Low Yes Yes Low Low Yes Yes Low Low Yes Yes No High
et al,4 2010
Gravel No U U U Yes Yes Low Low Yes Yes Low Low Yes Yes No High
et al,7
2010
Gorelick No Yes High High Yes U U Low Yes U U Low Yes Yes No High
et al,12
1997
Jauregui No No High High Yes No High Low U Yes U Low Yes Yes No High


et al,35

www.jpeds.com
2014
Mackenzie et al,37 U U U High Yes Yes Low Low Yes Yes Low Low Yes Yes Yes Low
1989
Parkin No U High Low U Yes U Low U U U Low Yes Yes Yes Low
et al,31
2010
Shavit et al,34 2006 No Yes High Low Yes Yes Low Low Yes Yes Low Low Yes Yes Yes Low
Steiner No Yes High High U U U Low U U U Low Yes Yes Yes Low
et al,11
2007
Vega et al,33 1997 Yes Yes Low High Yes Yes Low Low Yes Yes Low Low Yes Yes No High

QUADAS-2, Quality Assessment of Diagnostic Accuracy Studies 2; U, unclear.

*QUADAS-2 domain descriptions: 1.a. Random or consecutive sample; 1.b. Did the study avoid inappropriate exclusions?; 1.c. Was the study a low risk of bias?; 1.d.Is the study applicable?; 2.a. Reference standard results not known; 2.b. Prespecified threshold; 2.c.
Was the study a low risk of bias?; 2.d. Is the study applicable?; 3.a. Likely to classify target patients; 3.b. Index test results not known; 3.c. Was the study a low risk of bias?; 3.d. Is the study applicable?; 4.a. Interval between tests; 4.b. Same standard for all patients;
4.c. All patients included in analysis; 4.d. Was the study a low risk of bias?.

Vol. -, No. -
Freedman et al
- 2015 ORIGINAL ARTICLES

Table V. Description of measurement scales, scores, and technological devices used to assess dehydration
Study where scale/
Scale or score score is used Description Items
CDS Gravel 2010 4 items General appearance (0/1/2 points)
Jauregui 2014 Score range 0-8 points Eyes (0/1/2 points)
Parkin 2010 Grouped as 0 = no dehydration; 1-4 = some Mucous membranes (0/1/2 points)
dehydration; 5-8 moderate–to-severe dehydration Tears (0/1/2 points)

Physician’s clinical Chen 2010 Vega/Chen Vega/Chen

assessment Jauregui 2014 Mild: #5% General appearance
Mackenzie 1989 Moderate: 6%-9% Radial pulse
Shavit 2006 Severe: $10% Respirations
Vega 1997 Mackenzie Anterior fontanel
5%-7%: any clinical signs Blood pressure
7%-10%: obvious signs Skin turgor
>10%: circulatory failure Eyes
Shavit Tears
Overall clinical assessment that uses 7-point Likert scale Mucous membranes
Mackenzie
General appearance
Peripheral perfusion
Respirations
Anterior fontanelle
Urine output
Skin turgor
Eyes
Tears
Mucous membranes
Thirst
Shavit
Overall clinical assessment that
uses 7-point Likert scale

“Gorelick” score Gorelick 1997 4-Item Model General appearance

Jauregui 2014 Score range 0-4 points Radial pulse
$5% dehydration: 2 or more points Respirations
$10% dehydration: 3 or more points Skin turgor
10-Item Model Eyes
Score range 0-10 points Tears
$5% dehydration: 3 or more points Mucous membranes
$10% dehydration: 7 or more points Capillary refill
Heart rate
Urine output

Bedside ultrasound
Chen 2010 IVC and Ao diameters measured using C60 IVC:Ao using cut-off of 0.80 to
curvilinear probe over subxiphoid region indicate significant dehydration
Transverse image of vessels obtained
Maximal IVC diameter (expiratory phase)
Maximal Ao diameter (systole)

Digital capillary refill Shavit 2006 Hand raised slightly above level of heart DCRT $0.4 s
time (DCRT) Filmed fingertip using digital video camera, with
customized graphic software
“Frame-by-frame” analysis of exact color
Lightly pressed small rod against fingertip for 5 s and then released it
Compares subsequent frame with “precompression” frame until an
exact match is achieved (“recovery frame”)
Software calculates time between release and recovery frame = DCRT

Urine testing Steiner 2007 Visual calorimetric analysis at the bedside (performed by nurse or Urine ketones
physician) with Multistix – 10 SG urinalysis strips Urine specific gravity

IVC, Inferior Vena Cava; Ao, Aorta; DCRT, Digital Capillary Refill Time.

Diagnosing Clinically Significant Dehydration in Children with Acute Gastroenteritis Using Noninvasive Methods: 9.e4
A Meta-Analysis
THE JOURNAL OF PEDIATRICS  www.jpeds.com Vol. -, No. -

Figure 1. Flow diagram of study selection process. PICO, population intervention comparison outcome. From: Moher D, Liberati
A, Tetzlaff J, Altman DG, The PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA
statement. PLoS Med 2009; 6: e1000097. For more information, visit www.prisma-statement.org.

9.e5 Freedman et al
- 2015 ORIGINAL ARTICLES

Figure 2. QUADAS-2 risk of bias and applicability assessment, by domain. A, QUADAS-2 assessment of risk of bias by domain.
B, QUADAS-2 assessment of applicability by domain. QUADAS-2, Quality Assessment of Diagnostic Accuracy Studies 2.

Diagnosing Clinically Significant Dehydration in Children with Acute Gastroenteritis Using Noninvasive Methods: 9.e6
A Meta-Analysis