Professional Documents
Culture Documents
doi: 10.1093/tropej/fmy009
Original paper
A BS TR A C T
Background: Early detection of malnutrition in hospitalized children helps reduce length of hos-
pital stay and morbidity. A validated nutrition tool is essential to correctly identify children at risk of
malnutrition or who are already malnourished. This study compared the use of the Subjective
Global Nutrition Assessment (SGNA, nutrition assessment tool) and Screening Tool for the
Assessment of Malnutrition in Paediatrics (STAMP, nutrition screening tool) with objective nutri-
tional parameters to identify malnutrition in hospitalized children.
Methods: A cross-sectional study was carried out in two general paediatric wards in a public hos-
pital. SGNA and STAMP were performed on 82 children (52 boys and 30 girls) of age 1–7 years.
The scores from both methods were compared against Academy of Nutrition and Dietetics/
American Society of Parental and Enteral Nutrition Consensus Statement for identification of
paediatric malnutrition. The objective measurements include anthropometry (weight, height
and mid-arm circumference), dietary intake and biochemical markers (C-reactive protein, total
lymphocytes and serum albumin). Kappa agreement between methods, sensitivity, specificity and
cross-classification were computed.
Results: SGNA and STAMP identified 45% and 79% of the children to be at risk of malnutrition,
respectively. Using a compendium of objective parameters, 46% of the children were confirmed to
be malnourished. The agreement between SGNA and objective measurements (k ¼ 0.337) was
stronger than between STAMP and objective measurements (k ¼ 0.052) in evaluating the
nutritional status of hospitalized children. SGNA also has a 4-fold higher specificity (70.45%) than
STAMP (18.18%) in detecting children who are malnourished.
C The Author(s) [2018]. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
V 1
Conclusion: SGNA is a valid nutrition assessment tool in diagnosing malnutrition status among
hospitalized children in Malaysia. The discrepancy in specificity values between the two methods ex-
plains the distinguished roles between SGNA and STAMP. The use of STAMP will have to be fol-
lowed up with a more valid tool such as SGNA to verify the actual nutrition status of the paediatric
population.
and nutrition-related biochemical markers, including USA). Inadequate nutrient intake was considered
inflammatory markers. A single trained assessor eval- when a child consumed <75% of the estimated en-
uated each child for all tools. ergy intake. The present study used the definition of
malnutrition according to the Academy of Nutrition
Subjective Global Nutrition Assessment and Dietetics/American Society of Parental and
SGNA evaluates nutrition-focused medical history Enteral Nutrition (AND/ASPEN) [7]. Children are
(growth pattern, changes in body weight, adequacy identified as being malnourished if they experienced
of dietary intake, persistent gastrointestinal symp- abnormal anthropometric indicator supporting with
toms, functional impairment and metabolic stress) the presence of inadequate nutrient intake, inflamma-
and nutrition-focused physical examination (muscle tion or low albumin level [7].
wasting, fat loss and nutrition-related oedema)
for signs of inadequate energy and/or protein Statistical analysis
intake. Considering the absence or presence of Statistical analysis was performed using the SPSS
the nutrition-focused findings, a global rating of statistical software (Version 20.0, SPSS Inc., IBM
well-nourished, moderately malnourished or severely Corp). Data are presented as means 6 standard devi-
malnourished was established according to guide- ations, with statistical significance set at the 5% level.
lines provided on the SGNA rating form [16]. Cohen’s kappa and 95% confidence intervals were
Children with a Grade B or C under SGNA were calculated to test criterion and concurrent validity
classified as malnourished in the study. between the methods (SGNA/STAMP versus ob-
jective assessment). Children belonging to the mod-
Screening Tool for the Assessment of Malnutrition erately and severely malnourished groups were
Paediatrics consolidated in the presence of malnutrition risk cat-
STAMP consists of three questions evaluating the egory when calculating the diagnostic values (sensi-
medical condition, nutritional intake and anthro- tivity, specificity and positive and negative predictive
pometry of a child that yield a sum total to calculate values) of SGNA and STAMP.
the overall risk of malnutrition. Children with high
(score 4) or medium (score 2–3) risk of malnutri- Criterion validity
tion were classified as at risk of malnutrition [12]. The classification of malnutrition for each child using
the SGNA and STAMP was compared with a refer-
Objective nutritional parameters ence standard—a complete objective nutritional as-
Anthropometric measurements (height in centimetres sessment (according to AND/ASPEN guidelines)
using a stadiometer [SECA, Germany], weight in kilo- (n ¼ 82).
grams using a digital weighing scale [TANITA,
Japan], and mid-upper arm circumference in centi- Concurrent validity
metres using a measurement tape [SECA, Germany]) The child’s SGNA score was compared with the
were taken for all the children. Classification of an- STAMP score using the kappa statistic.
thropometric status was determined by using the
World Health Organization (WHO) Child Growth
RESULTS
Standard 2006 [5]. Malnutrition (serum albu-
min < 30 g/l) [17] or inflammatory (total lympho- Participants
cyte count > 33% [18] or C-reactive protein > 8 mg/ In total, 82 children (52 boys and 30 girls) partici-
l) [19] markers and medical histories were obtained pated in the study (Table 1). A majority of the par-
from the children’s medical records. A 24-h diet recall ticipants were of Malay ethnicity due to the normal
was performed to interview the parent/caregiver demographics of children in a public hospital
about the child’s dietary intake over the past 24 h. (Table 1) [20–21]. The length of hospital stay
Energy intake was analysed using the Nutritionist ranged from 1 to 13 days. A majority of the patients
ProTM Diet Analysis software (Axxya Systems LLC, were admitted due to respiratory conditions (45%).
Table 2. Cross-classification of nutritional status SGNA demonstrated its ability in identifying mal-
on SGNA, STAMP and objective assessment nourished children with good sensitivity (63.15%),
and it could correctly predict 64.86% of the children
No. of children (%)
who were truly malnourished. The performance of
Malnourished Well-nourished Total SGNA in this study differed from that reported by
Mahdavi et al., [10] which was 88.24% sensitivity,
Objective assessment
45.83% specificity and 60.61%positive predictive value
SGNA
among hospitalized Iranian children. In our study, to
Malnourished 24 (TP) 13 (FP) 37
diagnose a child with malnutrition, a child not only
Well-nourished 14 (FN) 31 (TN) 45
experienced subnormal anthropometric parameters
Total 38 44 82
but also an additional subnormal parameter, including
STAMP
inflammation markers, serum albumin or dietary in-
At risk of 29 (TP) 36 (FP) 65
take, according to AND/ASPEN consensus statement
malnutrition
[7]. Based on the study by Mahdavi et al., [10] chil-
Low risk of 9 (FN) 8 (TN) 17
dren were classified as malnourished when at least
malnutrition
two parameters (including weight-for-age, height-for-
Total 38 44 82
age, weight-for-height, triceps skinfold thickness and
SGNA
serum transferrin) were subnormal. The latter study
STAMP
might limit the diagnostic criteria to merely include
At risk of 28 (TP) 37 (FP) 65
the anthropometric variables for classifying those with
malnutrition
non-illness-related malnutrition. Thus, this discrep-
Low risk of 9 (FN) 8 (TN) 17
ancy might explain a higher false-positive error rate
malnutrition
(54.17%) observed in the other study when compared
Total 37 45 82
with this study (29.55%). Indeed, lack of a clear defin-
TP ¼ true positive; FP ¼ false positive; FN ¼ false negative; TN ¼ true
ition for paediatric malnutrition at present is a study
negative. limitation when comparing findings across countries.
Consistent with other studies, [12, 13, 23]
countries can be different due to the diversity in social STAMP offers high sensitivity as compared with
demographics, medical care services and food culture AND/ASPEN objective measures in detecting risk
[11, 22]. of malnutrition among children. Its high sensitivity
When comparing the agreement between SGNA produced high false-positive results (81.82%); there-
and objective methods, this study reported a moder- fore, well-nourished children experienced increased
ate-to-fair agreement that was consistent with other likelihood to be identified as being at high risk
studies [9–11] in identifying malnutrition status of of malnutrition using STAMP. On a side note, the
hospitalized children. The malnutrition evaluation cri- Consensus Statement of AND/ASPEN not purposes
teria in both SGNA and AND/ASPEN consensus to identify criteria for nutrition screening. McCarthy
statement include the assessment of anthropometry, et al. [12] introduced STAMP as a nutrition screen-
fluid accumulation, functional status and dietary ad- ing tool requiring no training and minimal time
equacy measures. Conversely, children are identified for identifying the likelihood of hospitalized children
as being malnourished by SGNA based on the clinical developing malnutrition. Although false-positive
judgement of healthcare professionals; thus, training results may call for detailed nutrition assessment on
on nutrition-focused physical examination in children the well-nourished children, it serves the purpose of
is required. In clinical settings, it might be impractical preventing nutritional status deterioration upon hos-
to include all variables in objective nutritional assess- pital admission. Thus, high sensitivity is ideal for a
ments; this study recommends that SGNA is a valid nutrition screening tool in identifying those children
yet comprehensive nutritional assessment tool to be at high risk of malnutrition during a prolonged
implemented in the local hospital settings. length of hospital stay.
Table 3. Statistical indicators between nutritional screening/assessment tools and objective assess-
ment in identifying children with malnutrition
Statistic indicators Formula SGNA STAMP
TP
Sensitivity (%) TPþFN 63.15 76.32
TN
Specificity (%) TNþFP 70.45 18.18
TP
Positive predictive value (%) TPþFP 64.86 44.62
TN
Negative predictive value (%) TNþFN 68.89 47.06
FP
False-positive error rate FPþTN 29.55 81.82
FN
False-negative error rate FNþTP 36.84 23.68
TPþTN
Accuracy (%) TPþFPþFNþTN 67.07 45.12
þ sensitivity
Likelihood ratio positive (LR ) 1specificity 2.14 0.93
1sensitivity
Likelihood ratio negative (LR ) specificty 0.52 1.33
LRþ
Odd ratio LR 4.12 0.70
TP ¼ true positive; FP ¼ false positive; FN ¼ false negative; TN ¼ true negative; LR ¼ likelihood ratio.
SGNA offers higher specificity (70.45%) than unequal gender distribution may be due to boys
STAMP (18.18%) when compared with malnutrition being more likely to be admitted to hospitals than
identification based on objective measurements girls, as reported in other local studies [20–21]. Our
as recommended by ASPEN/AND criteria. On study findings indicate that the boys experienced no
comparison with SGNA as the reference method significant difference in age, dietary intake and bio-
(concurrent validity determination), STAMP obtains markers, when compared with the girls. Thus, we
17.78% specificity. The discrepancy in specificity consolidated the participants to validate the applica-
values between the two methods explains their tion of SGNA in our local setting.
respective distinguished roles: SGNA as a nutrition
assessment tool requires more time and technical
skills to evaluate the nutritional status of children,
resulting in a higher accuracy (67.07%) in detect- CONCLUSION
ing malnourished children, whereas STAMP as a SGNA can be validly recommended for evaluating
nutrition screening tool aims to screen in a quick the malnutrition status of hospitalized children in
and easy manner children who are at high risk of Malaysian tertiary hospital settings. However, the
developing malnutrition. Hence, SGNA performs study found that STAMP cannot be recommended
better than STAMP in identifying well-nourished for use as a routine nutrition screening tool in local
children among the hospitalized children, which Malaysian children due to high chances of misclassi-
is reflected in its higher odds ratio (3.77) fication of their nutritional status. The use of
compared with STAMP (odds ratio ¼ 0.70). This STAMP will have to be followed up with a more
study also suggests that SGNA and STAMP should valid tool such as SGNA to verify the actual nutrition
not be used interchangeably in the clinical practice status of the paediatric population.
settings.
A higher number of boys and Malays participated
this study. The higher proportion of Malays in this ACKNOWLEDGEMENTS
study adequately represents the ethnic distribution The authors thank dieticians and nursing staffs from Hospital
in Malaysia (which comprises 50% Malays, 23% Selayang for supporting the conduct of research at its
Chinese and 7% Indians] [15]. However, the premise.