544321

research-article2014
PENXXX10.1177/0148607114544321Journal of Parenteral and Enteral NutritionWhite et al

Original Communication
Journal of Parenteral and Enteral
Nutrition
Simple Nutrition Screening Tool for Pediatric Inpatients Volume 40 Number 3
March 2016 392­–398
© 2014 American Society
for Parenteral and Enteral Nutrition
DOI: 10.1177/0148607114544321
jpen.sagepub.com
Melinda White, PhD1; Karen Lawson, BSc2; Rebecca Ramsey, PhD3; hosted at
Nicole Dennis, BHSc1; Zoe Hutchinson, BHSc4; Xin Ying Soh, BNutrDiet3; online.sagepub.com
Misa Matsuyama, BNutrDiet5; Annabel Doolan, BHSc6; Alwyn Todd, PhD7,8;
Aoife Elliott, BSc6; Kristie Bell, PhD1; and Robyn Littlewood, PhD1

Abstract
Background: Pediatric nutrition risk screening tools are not routinely implemented throughout many hospitals, despite prevalence studies
demonstrating malnutrition is common in hospitalized children. Existing tools lack the simplicity of those used to assess nutrition risk in the
adult population. This study reports the accuracy of a new, quick, and simple pediatric nutrition screening tool (PNST) designed to be used
for pediatric inpatients. Materials and Methods: The pediatric Subjective Global Nutrition Assessment (SGNA) and anthropometric measures
were used to develop and assess the validity of 4 simple nutrition screening questions comprising the PNST. Participants were pediatric
inpatients in 2 tertiary pediatric hospitals and 1 regional hospital. Results: Two affirmative answers to the PNST questions were found to
maximize the specificity and sensitivity to the pediatric SGNA and body mass index (BMI) z scores for malnutrition in 295 patients. The
PNST identified 37.6% of patients as being at nutrition risk, whereas the pediatric SGNA identified 34.2%. The sensitivity and specificity of
the PNST compared with the pediatric SGNA were 77.8% and 82.1%, respectively. The sensitivity of the PNST at detecting patients with a
BMI z score of less than −2 was 89.3%, and the specificity was 66.2%. Both the PNST and pediatric SGNA were relatively poor at detecting
patients who were stunted or overweight, with the sensitivity and specificity being less than 69%. Conclusion: The PNST provides a sensitive,
valid, and simpler alternative to existing pediatric nutrition screening tools such as Screening Tool for the Assessment of Malnutrition in
Pediatrics (STAMP), Screening Tool Risk on Nutritional status and Growth (STRONGkids), and Paediatric Yorkhill Malnutrition Score
(PYMS) to ensure the early detection of hospitalized children at nutrition risk. (JPEN J Parenter Enteral Nutr. 2016;40:392-398)

Keywords
pediatrics; life cycle; nutrition assessment; nutrition; administration; nutrition support practice

Clinical Relevancy Statement
This study presents a simple, validated nutrition screening tool
for use in the pediatric inpatient population to identify patients
at nutrition risk.
Introduction
Poor nutrition status of pediatric inpatients has negative
impacts on immune function, physical and cognitive develop-
ment, and clinical outcomes.1-4 It is essential to identify pediat-
ric inpatients at nutrition risk so that timely and appropriate
nutrition intervention and treatment plans may be implemented
and nutrition deterioration prevented to improve health
outcomes.
Nutrition screening of adult inpatients is a relatively well-
established practice, whereas routine nutrition screening of
pediatric inpatients is not commonplace. Several validated
pediatric nutrition risk screening tools described in the litera-
ture have been shown to be effective in identifying children at
risk of developing malnutrition.5-8 The nutrition screening cri-
teria of these tools in varying combinations and complex­
ities include nutrition intake, history of weight loss, medical
condition/diagnoses/pain, and anthropometry. All tools use a
scoring system. The Paediatric Nutrition Risk Score requires
the estimation of <50% of food intake and secondary reference
From the 1Department of Dietetics and Food Services, Royal Children’s
Hospital, Children’s Health Queensland Hospital and Health Service,
Herston, Queensland, Australia; 2Department of Nutrition and Dietetics,
Mater Children’s Hospital, South Brisbane, Queensland, Australia; 3School
of Exercise and Nutrition Sciences, Nutrition and Dietetics Program,
Queensland University of Technology, Kelvin Grove, Queensland, Australia;
4
Nutrition and Dietetics Department, Nambour Hospital, Nambour,
Queensland, Australia; 5School of Public Health, Griffith University,
Southport, Queensland, Australia; 6Department of Nutrition and Dietetics,
Mater Health Services, South Brisbane, Queensland, Australia; 7Mater Health
Services and Griffith Health Institute, Griffith University, Queensland,
Australia; and 8Raymond Terrace, South Brisbane, Queensland, Australia.
Financial disclosure: None declared.
Received for publication March 13, 2014; accepted for publication June
30, 2014.
This article originally appeared online on August 5, 2014.
Corresponding Author:
Melinda White, PhD, Department of Dietetics and Food Services, Royal
Children’s Hospital, Children’s Health Queensland Hospital and Health
Service, Lower Ground Floor, South Tower, Herston Rd, Herston, QLD
4029, Australia.
Email: Melinda.White@health.qld.gov.au
White et al 393
to tables to determine grades of pathologic conditions.7 Both
the Screening Tool for the Assessment of Malnutrition in
Pediatrics (STAMP) and the Paediatric Yorkhill Malnutrition
Score (PYMS) involve the use of anthropometric measures,
calculation of body mass index (BMI), and comparison to pre-
determined cutoffs.5,8 The Screening Tool Risk on Nutritional
status and Growth (STRONGkids) requires a subjective clini-
cal assessment and reference to a table with a list of high nutri-
tion risk disease.6 Unlike other screening tools, it has had a
secondary validation by an independent institution.9
The pediatric nutrition screening tools do not have the sim-
plicity of the nutrition screening tools used in adult nutrition
screening, such as the Malnutrition Screening Tool, which
consists of 2 simple questions and is widely used Australian
hospitals.5-8,10,11 The complexity of the existing pediatric nutri-
tion risk screening tools is a likely barrier to their routine
implementation.5-8 Ideally, a screening tool designed for pedi-
atric inpatients should have the simplicity of the nutrition
screening tools used for adults, but it is unknown if this is
possible in the pediatric inpatient population, who have the
nutrition intricacy of growth requirements. The criteria for a
nutrition screening tool include (1) a high degree of sensitiv-
ity, specificity, validity, and reliability; (2) simple and easy
implementation without the need for user training; (3) quick,
inexpensive, and noninvasive clinical utility; and (4) devel-
oped specifically for a pediatric inpatient population without
excluding infants and children with certain clinical diagno-
ses.10 The pediatric nutrition risk screening tools described in
the literature do not meet all of these criteria since each tool
contains various elements of nutrition assessment that gener-
ate a scoring system to trigger further nutrition assessment.
The utility of existing tools is therefore limited by the time
required to implement. In addition, certain existing tools are
not validated for use in infants, require clinical training to
carry out, and involve comparison of weight and/or height
with predetermined standards.5-8
The pediatric Subjective Global Nutritional Assessment
(SGNA) is a comprehensive nutrition assessment tool that has
been validated in preoperative surgical patients.12 It contains
several components of a nutrition-focused medical history and
physical examination and is the closest “gold-standard”
method available for assessing nutrition status in pediatric
patients.13 Although it is a comprehensive pediatric nutrition
assessment tool, it does not provide an option for a rapid and
simple pediatric nutrition screening tool due to the time
required for completion and its relative complexity.
The aim of this study is to produce a quick, simple, and
valid pediatric nutrition screening tool (PNST) to be used for
pediatric inpatients on admission to the hospital. The design
and use of the PNST should meet the previously outlined crite-
ria of simplicity and inclusively, with no user training require-
ments. Specifically, the PNST should avoid the need for
anthropometric measurement and a scoring system and have
the simplicity of the nutrition screening tools designed for
adults. Primarily, it should be effective in identifying pediatric
inpatients who are malnourished or at nutrition risk.
Methods
Patient Population
A convenience sample of children (n = 295) from term age to
16 years who were inpatients in 2 tertiary pediatric hospitals
and 1 regional hospital were recruited for the study from
September 2012 through June 2013. Corrected gestational age
was used for ex-premature infants born <37 weeks, up to 2
years chronological age. One of the tertiary hospitals had a par-
ticular interest in defining the nutrition status of its infant pop-
ulation and restricted data collection to infants. Patients were
excluded if they could not be weighed, were admitted for <24
hours, had conditions that markedly affected hydration, were
clinically unstable, or had non–English-speaking parents or
caregivers. Patient demographic data collected included age,
sex, primary diagnosis, reason for admission, and length of
stay at the time of measurement.
Ethical approval was obtained from the Children’s Health
Services Human Research Ethics Committee (approval num-
ber HREC/12/QRCH/107). Site-specific approval was obtained
for each hospital. The study was conducted in full accordance
with the Australian Code for the Responsible Conduct of
Research, the National Health and Medical Research Councils
National Statement on Ethical Conduct in Human Research
(2007). All participants were recruited after receiving written
information and a verbal explanation of the study and obtain-
ing written consent from parents or caregivers.
Anthropometry, PNST, and Pediatric SGNA
Anthropometry, the PNST, and the pediatric SGNA were com-
pleted for each patient on the same day by the same investiga-
tor. To eliminate bias, the PNST questions were asked before
any other measurements were taken. Weight, height, or length
was measured using methods previously described by the
World Health Organization (WHO).14 Each site sought the
calibration of all scales and stadiometers prior to study com-
mencement. The anthropometric standards used were recom-
mended by the Centers for Disease Control and Prevention
(CDC). BMI-for-age, height-for-age, and weight-for-age z
scores were calculated using WHO growth standards for 0–2
years of age (www.who.int/childgrowth/standards/en) and the
U.S. CDC 2000 reference population for 2–20 years of age
(www.cdc.gov). In addition, CDC reference population data
were chosen for the 2- to 20-year age group since WHO weight
curves extend only to 10 years of age. Malnutrition, wasting,
and stunting were defined as ≤−2 z score for BMI-for-age,
weight-for-age, and height-for-age, respectively.14 Obesity was
defined according to the internationally agreed standard cutoff
point for overweight and obese, where overweight is defined as
394 Journal of Parenteral and Enteral Nutrition 40(3)
Table 1.  Nutrition Screening Questions Incorporated Into the
Pediatric Nutrition Screening Tool.
No. Question
1      Has child unintentionally lost weight lately?
2      Has child had poor weight gain over the last few months?
3      Has child been eating/feeding less in the last few weeks?
4     Is child obviously underweight/significantly overweight?
at or above the 85th percentile and lower than the 95th percen-
tile and obesity is defined as at or above the 95th percentile.15
Development of the PNST
The nutrition screening questions included in the PNST were
selected via a thorough review of the literature of existing screen-
ing tools and through consultation with nursing and dietetic staff
at a tertiary pediatric hospital. Nurses (n = 3) involved in devel-
oping a hospital admission form had input into the PNST ques-
tions, and senior clinical pediatric dietitians (n = 5) were asked
for feedback. Further comment was sought by discussing the
questions at a Nurses Unit Manager meeting where more than 10
nurses attended. A primary contributing factor to the choice of
questions was ease of application and that they used minimal
space on an admission form, which had several other components
related to the patient’s clinical condition. The PNST consisted of
4 simple questions that required a yes or no response (Table 1).
Based on clinical judgment, 2 affirmative responses were chosen
as a predictor of nutrition risk by the PNST. The pediatric SGNA
and anthropometrics were chosen as the gold standards in defin-
ing nutrition risk and were completed using previously described
methods.13,14 The pediatric SGNA was compared with the PNST,
and anthropometry was compared with the pediatric SGNA and
the PNST. A patient with an overall pediatric SGNA rating of
moderate or severe was considered at nutrition risk.
Statistical Analysis
Data were analyzed using jMetrik freeware (JP. Meyer,
University of Virginia, USA), SPSS version 20 (SPSS, Inc, an
IBM Company, Chicago, IL), and R software (R Foundation
for Statistical Computing, Vienna, Austria). Descriptive statis-
tics were used for the presentation of demographic and anthro-
pometric patient characteristics.
Development of the PNST.  To determine the individual con-
tribution of each PNST question toward identifying children
who were malnourished or at nutrition risk, we undertook a
principal components factor analysis. Rasch analyses gener-
ated INFIT values by using jMetrik freeware to identify the
discriminative capacity of each question by assessing the
expected responses compared with actual responses. Fit sta-
tistics were generated to investigate how well each item and
participant conformed to the assumptions of the model. Item-
INFIT, the most commonly used type of fit statistic, is an
information-weighted statistic that compares the discrimina-
tion of an individual item with the average discrimination of
all items in a measure. INFIT values investigate whether
items in a scale measure the same construct independently of
each other.16 INFIT values below 1 indicate fewer affirmative
responses than expected, suggesting an item is redundant.
INFIT values above 1 suggest that more people answered
affirmatively than predicted by the model and that the infor-
mation contributed to by the item may be “overvalued.” Cut-
off values were set at a recommended range of 0.8–1.2, with
a wider acceptable range of 0.7–1.3.17
Cumulative percentages for number of affirmative responses
were used to identify the threshold at which the PNST identi-
fied a proportion of children at nutrition risk as determined by
the pediatric SGNA.
Validation of the PNST.  Based on the identified cutoff points
for malnutrition and overweight using z scores and nutrition
risk (pediatric SGNA), the sensitivity, specificity, positive pre-
dictive value, and negative predictive value were identified for
the PNST. The z scores were also compared with the sensitiv-
ity, specificity, positive predictive value, and negative predic-
tive value of the pediatric SGNA. The McNemar test was used
to investigate the difference in the proportion of children iden-
tified as at nutrition risk by the PNST compared with the pedi-
atric SGNA. Correlations used to determine validity between
the PNST, z scores, and the pediatric SGNA were investigated
via polychoric correlation using R.
Results
Patients and Anthropometry
Table 2 summarizes the characteristics of the pediatric inpa-
tients (n = 295) included in the study. Thirty-one patients were
recruited from the pediatric ward in the regional hospital, 154
patients from 1 tertiary pediatric hospital, and 110 from the
tertiary hospital that recruited only patients up to age 12
months. Therefore, the median (interquartile range) age of the
total population was relatively young at 10.9 (4.3–70.0)
months, with 155 patients younger than 12 months. The pri-
mary reason for admission was respiratory illness, and more
than 50% of the primary diagnoses fell into the “other” cate-
gory. Examples of “other” primary diagnoses included respira-
tory infections, trauma, head injury, or cerebral palsy. The
median (interquartile range) weight, height or length, BMI,
and BMI percentile were 10 (6.4–19.7) kg, 75 (64–111) cm,
16.48 (14.65–17.8) kg/m2, and 40th percentile (<3rd, >97th),
respectively. The median (interquartile range) weight-for-age z
score was −0.03 (−0.87 to 0.91), height-for-age z score was
0.31 (−0.64 to 1.29), and BMI z score was −0.26 (−1.26 to
0.87). Eighty-two (27.8%) patients were overweight or obese.
White et al 395

Table 2.  Characteristics of Pediatric Inpatients (n = 295).

Characteristic Infants <12 mo (n = 155) Children ≥12 mo (n = 140)

Sex
 Male 97 (62.6) 69 (49.3)
 Female 58 (37.4) 71 (50.7)
Age, median (interquartile range), mo 4.5 (2.15–7.68) 73.0 (38.0–116.0)
Length of stay, median (interquartile range), days 2.0 (2.0–5.0) 3.0 (2.0–8.0)
Primary diagnosis
  Oncology, active treatment 2 (1.3) 31 (22.1)
  Cystic fibrosis 0 18 (12.9)
 Gastroenterology 27 (17.4) 15 (10.7)
 Surgical 10 (6.5) 17 (12.1)
 Cardiac 15 (9.7) 0
 Other 101 (65.2) 58 (41.4)
Reason for admission
 Surgery 21 (13.5) 33 (23.6)
 Gastroenterology 10 (6.5) 4 (2.9)
 Infection 36 (23.2) 25 (17.9)
  Respiratory infection/illness 59 (38.1) 32 (22.9)
 Other 29 (18.7) 46 (32.9)

Values are presented as number (%) unless otherwise indicated.

Table 3.  Affirmative Responses to Individual Questions, Dimensions, and Component Loading From Factor Analysis and INFIT
Values From Rasch Analyses of the Pediatric Nutrition Screening Tool.

No. (%) of Dimension

Question Affirmative Responses (Component Loading) INFIT Value
1. Has child unintentionally lost weight lately? 89 (30.2) 1 (0.733) 0.84
2. Has child had poor weight gain over the last few months? 80 (27.1) 1 (0.777) 0.82
3. Has child been eating/feeding less in the last few weeks? 154 (52.2) 1 (0.577) 1.00
  2 (−0.564)  
4. Is child obviously underweight/significantly overweight? 56 (19.0) 2 (0.720) 1.16

Development of the PNST Table 4.  Cumulative Percentage of Affirmative Responses

Affirmative responses to individual questions, dimensions, and
component loading from the factor analysis and Rasch analysis
are summarized in Table 3. Factor analysis revealed all ques-
tions loaded highly onto an underlying dimension, and all
items fell within the widely accepted range for fit, resulting in
all the questions being equally important for inclusion into the
PNST. The cumulative percentage of affirmative responses
from the PNST that agreed with the prevalence of patients
identified as at nutrition risk by the pediatric SGNA is shown
in Table 4. Based on the cumulative percentage of affirmative
responses to the PNST, a threshold of 2 affirmative responses
provided a proportion of children at nutrition risk closely
identified by the pediatric SGNA.
Validation of the PNST
Using the cutoff point of 2 affirmative responses, the
PNST identified 37.6% of patients as at nutrition risk, and the
From the Pediatric Nutrition Screening Tool That Agreed With
the Prevalence of Patients Identified as at Nutrition Risk by the
Pediatric Subjective Global Nutrition Assessment.
Number of Affirmative Cumulative
Responses Percentage
0 29.8
1 61.6
2 82.7
3 95.2
4 100
pediatric SGNA identified 34.2% as at nutrition risk. The per-
centage of patients who met the definition for malnutrition of
≤−2 BMI z scores was 9.8%, and 2% of these patients had
severe malnutrition with a BMI z score ≤−3. Table 5 shows the
sensitivity, specificity, positive predictive values, and nega-
tive predictive values identified for the PNST compared with
396 Journal of Parenteral and Enteral Nutrition 40(3)

Table 5.  Sensitivity, Specificity, and Predictive Values for the PNST vs z Score Cutoffs and the Pediatric SGNA and the Pediatric
SGNA vs z Score Cutoffs to Identify Malnourished and Overweight Patients.

Risk Identified by Pediatric SGNA Risk Identified by PNST

Measure No. (%) Sensitivity, % Specificity, % PPV, % NPV, % Sensitivity, % Specificity, % PPV, % NPV, %
PNST (≥2 affirmative 111 (37.6) NA NA NA NA NA NA NA NA
responses)
Weight-for-age z score
  ≤−2 21 (7.1) 85.7 69.7 17.8 98.5 89.5 65.0 15.3 98.9
  ≤−3 7 (2.4) 100.0 67.4 6.9 100.0 100.0 62.9 5.4 100.0
Height-for-age z score
  ≤−2 13 (4.4) 46.2 66.5 6.0 96.4 55.6 62.4 4.5 97.8
  ≤−3 5 (1.7) 60.0 66.4 3.0 99.0 75.0 62.3 2.7 99.4
BMI z score
  ≤−2 29 (9.8) 96.5 72.5 27.7 99.5 89.3 66.2 22.5 98.4
  ≤−3 6 (2.0) 100.0 67.0 59.4 100.0 100.0 62.8 5.4 100.0
  ≥85th percentile 82 (27.8) 42.5 68.4 33.7 75.1 45.5 64.1 32.4 75.8
Pediatric SGNA 101 (34.2) NA NA NA NA 77.8 82.1 69.3 87.6

BMI, body mass index; NA, not applicable; NPV, negative predictive value; PNST, Pediatric Nutrition Screening Tool; PPV, positive predictive value;
SGNA, Subjective Global Nutrition Assessment.

z score cutoffs and the pediatric SGNA and for the pediatric Table 6.  Correlation Between the Nutrition Risk Identified by
SGNA compared with z score cutoffs. The sensitivity and the PNST With the Nutrition Risk Identified by the Pediatric
specificity for the PNST compared with the pediatric SGNA SGNA and Nutrition Status by z Scores.
were moderate and high, scoring 77.8% and 82.1%, respec- Correlation Correlation
tively. The sensitivity of the PNST at detecting patients with Measure PNST Pediatric SGNA
malnutrition was higher at 89.3%, but the specificity decreased
Weight-for-age z score
to 66.2%. The PNST and pediatric SGNA detected 100% of
  ≤−2 −0.66 −0.66
patients with severe malnutrition. Both the PNST and pediat-
  ≤−3 −0.93 −0.95
ric SGNA were relatively poor at detecting patients who were
Height-for-age z score
stunted or overweight, with a sensitivity and specificity <67%.
  ≤−2 −0.19 −0.15
Finally, validation was demonstrated with the PNST being   ≤−3 −0.37 −0.27
highly correlated with the pediatric SGNA, and correlations BMI z score
between the PNST and z scores, as well as the pediatric SGNA   ≤−2 −0.70 −0.85
and z scores, were also comparable (Table 6).   ≤−3 −0.93 −0.94
  ≥85th percentile −0.15 −0.16
Discussion Pediatric SGNA 0.80 1.00

It has previously been identified that “a simple and reliable
nutrition risk screening tool appears highly desirable to
advance the early and cost-effective identification of children
who will benefit from targeted nutrition intervention.”18,p.304
This study resulted in the production of a simple and effective
nutrition risk screening tool for the early detection of at-risk
hospitalized children requiring more thorough nutrition assess-
ment and individualized nutrition intervention.
During the design of the PNST, we deliberately aimed to
avoid the inclusion of anthropometric assessment using weight
and/or height measures. Ideally, weight and height measures
should be part of routine admission of pediatric inpatients and
would enhance the nutrition screening process. Although
weight is usually part of a routine hospital admission process,
often height or length is not measured, and more in-depth
BMI, body mass index; PNST, Pediatric Nutrition Screening Tool;
SGNA, Subjective Global Nutrition Assessment.
nutrition assessment is required to compare weight and height
with growth standards. The frequency of comparison to growth
standards should improve since more hospitals are starting to
use electronic tools to calculate z scores and plot anthropomet-
ric measures on growth charts. Although calibrated and func-
tional anthropometric equipment should be part of standard
practice in pediatric healthcare institutions, it can often be
lacking, and measurement techniques can be poor.19,20 During
the PNST design, nursing staff expressed a reluctance to com-
pare patients’ weights and heights with any form of growth
standard, including BMI cutoff tables as used in PYMS and
STAMP or percentile charts. Consideration also had to be
White et al 397
made to a space restriction around the size of the tool if it was
to be included in the admission forms, and the wording was
required to be brief. Therefore, to replace anthropometric mea-
sures, a subjective question of “Is child obviously underweight/
significantly overweight?” was included in the PNST.
The screening criteria of a “high nutrition risk disease” are
commonly used in the existing pediatric nutrition screening
tools. This criteria were omitted from our tool to avoid the refer-
ence to a list of diagnoses. In addition, it was difficult to catego-
rize diagnoses, as demonstrated by more than 50% of children
being included in the “other” category for diagnosis (Table 2).
Decreased nutrition intake is a major contributor to malnu-
trition in hospitalized pediatric inpatients and has been incorpo-
rated into most of the existing pediatric nutrition screening
tools.5,6,8 We found that more than 50% of patients had an affir-
mative response to “Has child been eating/feeding less in the
last few weeks?” indicating decreased oral or enteral intake as a
major nutrition risk factor for hospitalized children. Decreased
nutrition intake corresponds to weight loss, and up to 65% of
hospitalized children have been observed to lose weight during
admission.7 This is concerning since children should gain
weight as they grow, and impaired weight gain is associated
with poor clinical outcome. Therefore, a question about “poor
weight gain over the last few months” was included as part of
the PNST criteria.21 This question is subjective, and there is the
possibility of bias in the interpretation of the question since it
gives a loose time frame of a “few months,” but it did contribute
equally to the agreement with the pediatric SGNA and therefore
was kept in the PNST.
This study found that all the nutrition risk factors investi-
gated (ie, weight loss, poor weight gain, decreased nutrition
intake, and a subjective question of “obviously underweight/
significantly overweight”) were all equally important for inclu-
sion in the PNST. Children who are overweight or obese also
have an increased risk of mortality and morbidity when unwell.4
The prevalence of overweight and obesity in Australian pediat-
ric inpatients has been reported to be between 22% and 25%.22,23
Given the high prevalence of overweight and obese pediatric
inpatients, the aim of the PNST was to identify not only chil-
dren already malnourished or at risk of malnutrition but also
those who were overweight or obese. The specificity and sensi-
tivity of the PNST to detect children who were overweight and
obese were poor, and therefore other screening methods should
be employed to detect this population of inpatients.
This study used the pediatric SGNA as a gold standard to
test the validity of the PNST. The pediatric SGNA was also
chosen to assess the validity of the PYMS screening tool in a
previous study.8 Another study used the pediatric SGNA to
compare its performance with objective assessment, which
included anthropometric assessment, and found relatively poor
agreement.24 Conversely, we found that the pediatric SGNA
had a moderate specificity and high sensitivity in identifying
malnourished children who had a BMI z score ≤−2.
Two affirmative questions to the PNST were found to maxi-
mize the specificity and sensitivity to the pediatric SGNA and
BMI z scores for malnutrition. The threshold of 2 affirmative
questions was initially selected for testing based on clinical
judgment. High levels of sensitivity were found between the
PNST and the pediatric SGNA and the BMI z scores for mal-
nutrition. Reassuringly, the PNST identified 100% of children
with severe malnutrition. Due to this high level of agreement,
the PNST could play a role in identifying children who fit the
Diagnosis-Related Groups (DRG) reimbursement criteria for
malnutrition.25 Both the PNST and pediatric SGNA were poor
at identifying children who had a low height-for-age z score,
indicating a poor ability to detect children with chronic malnu-
trition. However, they were both able to identify children with
a reduced weight-for-age z score, indicating they are better at
detecting children with acute malnutrition.
This study did not collect interrater reliability or reproducibil-
ity data, which have a role in the development of a nutrition
screening tool. However, given that the PNST questions are self-
explanatory, involve no clinical skill to deliver, and only require a
negative or positive response, it is highly likely that interrater
agreement and reproducibility would have been acceptable, but
further studies are required to confirm this. The PNST is primarily
designed for screening on admission, but the validation studies
occurred while patients were at varying parts of their stay. It is
important that pediatric patients are rescreened throughout their
admission since children are at high risk of developing malnutri-
tion during admission, which has been shown to increase length of
stay and the likelihood of readmission.26 The PNST may have lim-
ited utility in children who are overweight or those with chronic
malnutrition due to reliance on visual inspection rather than
anthropometric measures.
In conclusion, the PNST provides a valid and simple alter-
native to existing pediatric-specific nutrition screening tools
designed for the pediatric inpatient population. It has applica-
tion for a wide variety of clinical diagnoses and age groups,
including infants, and is designed to identify children who
require a more in-depth nutrition assessment. Scope remains to
conduct a prospective follow-up study investigating the out-
comes of infants and children identified as being at nutrition
risk (via the PNST) and receiving subsequent targeted nutrition
intervention. There is also possibility for the modification of
the PNST to improve the detection of children who are signifi-
cantly overweight. Further studies are required to provide
independent validation of the PNST.
Acknowledgments
Our thanks go to all the participating children and their parents for
their cooperation, as well as all the participating hospitals. We
thank Kathy Beck from the Royal Children’s Hospital, Brisbane
and Mahalia Pappas from Queensland University of Technology
for their contribution in initiating the study.
Statement of Authorship
M. White, K. Lawson, A. Todd, A. Doolan, A. Elliott, K. Bell, and
R. Littlewood contributed to conception/design of the research; M.
398 Journal of Parenteral and Enteral Nutrition 40(3)
White, R. Ramsey, N. Dennis, Z. Hutchinson, X. Y. Soh, and M.
Matsuyama contributed to acquisition, analysis, or interpretation
of the data; and M. S. White drafted the manuscript. All authors
critically revised the manuscript, agreed to be fully accountable for
ensuring the integrity and accuracy of the work, and read and
approved the final manuscript.
References
1. Cunningham-Rundles S, McNeeley DF, Moon A. Mechanisms of
nutrient modulation of the immune response. J Allergy Clin Immunol.
2005;115(6):1119-1128.
2. Cao J, Peng L, Li R, et al. Nutritional risk screening and its clinical signifi-
cance in hospitalized children. Clin Nutr. 2014;33(3):432-436.
3. Corbett SS, Drewett RF. To what extent is failure to thrive in infancy asso-
ciated with poorer cognitive development? A review and meta-analysis. J
Child Psychol Psychiatry. 2004;45(3):641-654.
4. White M, Murphy A, Hallahan A, Ware R, Fraser C, Davies PSW.
Survival in overweight and underweight children undergoing bone mar-
row transplantation. Eur J Clin Nutr. 2012;66:1120-1123.
5. McCarthy H, Dixon M, Crabtree I, Eaton-Evans M, McNulty H. The
development and evaluation of the Screening Tool for Assessment of
Malnutrition in Paediatrics (STAMP) for use by healthcare staff. J Hum
Nutr Diet. 2012;25:311-318.
6. Hulst JM, Zwart H, Hop WC, Joosten KFM. Dutch national survey to test
the STRONGkids nutritional risk screening tool in hospitalized children.
Clin Nutr. 2010;29(1):106-111.
7. Sermet-Gaudelus I, Poisson-Saloman A, Colomb V, et al. Simple pediatric
nutritional risk score to identify children at risk of malnutrition. Am J Clin
Nutr. 2000;72:64-70.
8. Gerasimidis K, Keane O, Macleod I, Flynn DM, Wright CM. A four stage eval-
uation of the Paediatric Yorkhill Malnutrition Score in a tertiary paediatric
hospital and a district general hospital. Br J Nutr. 2010;104(5):751-756.
9. Huysentruyt K, Alliet P, Muyshont L. The STRONGkids nutritional
screening tool in hospitalized children: a validation study. Nutrition.
2013;26(11-12):1356-1361.
10. Ferguson M, Capra S, Bauer J, Banks M. Development of a valid and
reliable malnutrition screening tool for adult and acute hospital patients.
Nutrition. 1999;15(6):458-464.
11. Agarwal E, Ferguson M, Banks M, et al. Nutrition care practices in hos-
pital wards: results from the Nutrition Care Day Survey 2010. Clin Nutr.
2012;31(6):995-1001.
12. Secker DJ, Jeejeebhoy KN. How to perform Subjective Global
Nutritional Assessment in children. J Acad Nutr Diet. 2012;112(3):424-
431.
13. Secker DJ, Jeejeebhoy KN. Subjective Global Nutritional Assessment for
children. Am J Clin Nutr. 2007;85:1083-1089.
14. World Health Organization. Physical Status: The Use and Interpreta­
tion of Anthropometry. Report of a WHO Expert Committee. Geneva,
Switzerland: World Health Organization; 1995. WHO Technical Report
Series 854.
15. Cole TJ, Bellizi MC, Flegal KM, Dietz WH. Establishing a standard defi-
nition for child overweight and obesity worldwide: international survey.
Br Med J. 2000;320:1240-1243.
16. Smith RM, Schmacker RE, Bush MJ. Using item mean squares to evaluate
fit to the Rasch model. J Outcome Measurements. 1998;2:66-78.
17. Wright BD, Linacre JM, Gustafson JE, Martin-Löf P. Reasonable mean-
square fit values. Rasch Measurement Trans. 1994;8(3):370.
18. Hartman C, Shamir R, Hecht C, Koletzko B. Malnutrition screen-
ing tools for hospitalized children. Curr Opin Clin Nutr Metab Care.
2012;15(3):303-309.
19. Hutteman M, van der Ende J, Schweizer JJ. Presence and the function-
ing of scales and stadiometers in paediatric units. Clin Nutr. 2008;27:
171-172.
20. Bunting J, Waver LT. Anthropometry in a children’s hospital: a study
of staff knowledge, use and quality of equipment. J Hum Nutr Diet.
1997;10:17-23.
21. Eskedal LT, Hagemo PS, Seem E, et al. Impaired weight gain predicts risk
of late death after surgery for congenital heart defects. Arch Dis Child.
2008;93(6):495-501.
22. Aurangzeb B, Whitten KE, Harrison B, et al. Prevalence of malnutri-
tion and risk of under-nutrition in hospitalised children. Clin Nutr.
2012;31(1):35-40.
23. O’Connor J, Youde LS, Allen JR, Baur LA. Obesity and under-nutri-
tion in a tertiary paediatric hospital. J Paediatr Child Health. 2004;40:
299-304.
24. Mahdavi AM, Safaiyan A, Ostadrahimi A. Subjective vs objective nutri-
tional assessment study in children: a cross-sectional study in the north-
west of Iran. Nutr Res. 2009;29(4):269-274.
25. Commonwealth Department of Health and Aged Care. Australian Refined
Diagnosis Related Groups, Version 4.1, Definitions Manual. Vols 1-3.
Canberra, Australia: Commonwealth Department of Health and Aged
Care; 1998.
26. Rocha GA, Rocha EJM, Martins CV. The effects of hospitalization on the
nutritional status of children. J Pediatr (Rio J). 2006;82(1):70-74.