Clinical Nutrition 31 (2012) 206e211

Contents lists available at SciVerse ScienceDirect

Clinical Nutrition
journal homepage: http://www.elsevier.com/locate/clnu

Original article

Nutritional risk screening in surgery: Valid, feasible, easy!

Ana Isabel Almeida, Marta Correia, Maria Camilo, Paula Ravasco*
Unidade de Nutrição e Metabolismo, Instituto de Medicina Molecular, Faculdade de Medicina de Lisboa, Avenida Prof. Egas Moniz, 1649-028 Lisboa, Portugal

a r t i c l e i n f o s u m m a r y

Article history: Background & aims: We aimed to test the capacity in identifying patients at nutritional risk, by comparing
Received 4 August 2011 BMI, recent %weight loss,Nutritional Risk Screening 2002(NRS-2002),Malnutrition Universal Screening
Accepted 11 October 2011 Tool(MUST) and Nutritional Risk Index(NRI) with Subjective Global Assessment(SGA),considered the
Standard. The main purpose was to select the most consistent screening method for effective integration
Keywords: in daily surgical wards’ practice.
Nutritional risk
Methods: 300 surgical patients were assessed at admission: BMI(categorized by WHO’s criteria), weight
Nutritional screening
loss
5% in previous 6 months, NRS-2002, MUST, NRI, SGA. Concordances, correlations, sensitivity,
Undernutrition
Weight loss
specificity, positive(PPV) and negative predictive values(NPV) were calculated to evaluate methods’
Surgical patients performance vs the Standard.
Results: Prevalence of nutritional risk was 66% by NRS-2002 þ MUST, and 87% by NRI. By SGA, 64%
patients were undernourished. All methods agreed with SGA(k ¼ 0.85e0.91,p < 0.001), except BMI &
NRI(k ¼ 0.07e0.34,p < 0.05). NRS-2002, MUST and %weight loss effectively detected patients at risk:
sensitivity 0.8e0.89, specificity 0.89e0.93, PPV 81%e89%, NPV 89%e100%. Conversely, BMI & NRI were
ineffective: sensitivity 0.29e0.43, specificity 0.27e0.39, PPV 24%e35%, NPV 27%e31%; %weight loss alone
vs MUST/NRS-2002 was explored: sensitivity 0.79e0.87, specificity 0.85e0.89, PPV 84%e85%, NPV 87%
e89%, thus successfully identifying undernutrition risk.
Conclusions: In surgical patients, MUST þ NRS-2002 are valid for nutritional screening; recent weight loss

5% also proved highly efficient; its easy/quick calculation may facilitate adherence/integration by health
professionals as a minimum obligatory in clinical practice.
Ó 2011 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

1. Introduction consensus on which should be used as the gold standard to detect

Almost 80 years have elapsed since the relationship between
preoperative weight loss and postoperative complications was first
documented.1 Despite the recognized clinical significance, disease-
related undernutrition continues to be a common finding in 20%e
50% of hospitalized patients, depending on the population and
criteria used.2e4 The prevalence of undernutrition is apparently
higher amongst surgical patients, ranging from 35% to nearly
60%.5e9
Undernutrition has been consistently associated with poorer
clinical outcomes, e.g. impaired wound healing, higher infection
rates and mortality, longer length of stay, bearing to increased
overall costs.10,11 Yet, undernutrition is potentially reversible with
appropriate nutritional support, thereby early identification of high
risk patients is crucial for patient’ centered quality of care.12 Despite
the multitude of nutritional screening tools, there still is no
* Corresponding author. Tel.: þ351 217985141; fax: þ351 217985142.
E-mail address: p.ravasco@fm.ul.pt (P. Ravasco).
hospitalized patients at risk of undernutrition.13e15
The aim of this prospective cross-sectional study conducted in
hospitalized surgical patients was to select the most consistent
method for integration in everyday practice. For this purpose,
individual results of Body Mass Index (BMI), %weight loss in the
previous 6 months, and of nutritional screening tools such as
Nutritional Risk Screening (NRS-2002), Malnutrition Universal
Screening Tool (MUST) and Nutritional Risk Index (NRI), were
compared with Subjective Global Assessment (SGA).
2. Material and methods
2.1. Study population
This prospective study was carried out at a single large general
surgical department of a University Hospital in Lisbon, Portugal.
During 8 months, all consecutive adult patients (
18 years) with
a predicted length of stay
4 days, were considered eligible.
Exclusion criteria comprised organ transplantation, coma,

0261-5614/$ e see front matter Ó 2011 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
doi:10.1016/j.clnu.2011.10.003
 A.I. Almeida et al. / Clinical Nutrition 31 (2012) 206e211 207

bedridden, intensive care patients, or previous surgery, chemo/
radiotherapy during the year prior to hospital admission; patients
submitted to surgery prior to nutritional assessment were also
excluded. The study was approved by the University Hospital Ethics
Committee and was conducted in accordance with the Helsinki
Declaration of 1975 as revised in 1983; participant patients gave
their written informed consent.
2.2. Study measures
All nutritional data were collected by a single research dietitian
(MC). Nutritional risk and status were always assessed prior to any
surgical procedure and within three days after hospital admission.
2.3. Nutritional status
Body Mass Index. Height was measured in the upright position
using a stadiometer and weight was determined with a SecaÒ floor
scale and registered to the nearest 0.5 kg. Height and weight were
used to determine Body Mass Index [BMI ¼ weight(kg)/
height(m)2]. BMI was classified using the World Health Organiza-
tion (WHO) criteria: underweight if <18.5 kg/m2, adequate if
18.5e24.9 kg/m2, overweight if 25e29.9 kg/m2 or obese if
30 kg/
m2.16
Percentage of weight loss before admission was calculated and
classified as clinically significant and suggestive of undernutrition
or of nutritional risk, whenever
5% over the last 6 months.17
Subjective Global Assessment (SGA) is based on the patient’s
clinical history, physical examination, estimated weight loss and
changes in diet intake allowing the categorization of nutritional
status as: A) well nourished, B) moderately undernourished or
suspected of being undernourished, or C) severely undernour-
ished.18 SGA was chosen as the reference method.9,19 In order to
enable methods comparisons, SGA results were grouped into 2
categories: well nourished and moderately þ severely
undernourished.
2.4. Nutritional risk
Nutritional Risk Screening-2002 (NRS-2002) screens patients by
assessing BMI, unintentional weight loss, changes in food intake
and a disease severity score (low, moderate or severe), further
adjusted for age
70 years. The patient is at nutritional risk when
the final end result is
3.13
Malnutrition Universal Screening Tool (MUST) integrates BMI,
unintentional weight loss and scores the acute disease effect on
intake. Nutritional risk is categorized as follows: 0 ¼ low risk,
1 ¼ medium risk and
2 ¼ high risk.14
Nutritional Risk Index (NRI) was specifically devised for surgical
patients and relies on serum albumin concentration and the ratio of
actual to usual weight; it is defined by the formula [1.519  serum
albumin (g/L) þ 0.417  (actual weight/usual weight)  100]. A
score >100 indicates no nutritional risk, >97.5e100 a low risk,
83.5e97.5 a medium risk and a high nutritional risk if <83.5.15 NRI
could only be calculated in patients with available plasma albumin
values at the time of nutritional evaluation (n ¼ 237).
For the present study in order to enable methods comparisons,
two-categories of MUST and NRI results were created: low nutri-
tional risk and medium þ high nutritional risk.
2.5. Statistical analysis
Data was analyzed using SPSS 14.0 statistical software (SPSS Inc,
Chicago, USA). For every diagnostic group, all multivariate analyses
were conducted with the adequate adjustments for age and sex.
Categorical data were expressed as number of patients and
percentage; continuous data with a normal distribution were
expressed as mean  standard deviation, while continuous data not
normally distributed were expressed as median (range). Compari-
sons were made using c2 test, Student’s t-test or non-parametric
tests as appropriate. Agreement analysis between screening and
assessment methods was performed using Kappa and Spearman
correlation coefficients. The Kappa statistic was further classified
according to Fleiss.20
Sensitivity, specificity and predictive values were calculated to
appraise the comparative value of nutritional screening tools.
Sensitivity was expressed as the proportion of “undernourished”
cases classified by SGA, which were correctly classified as “at
nutritional risk” by the screening tools. If a screening tool has a high
sensitivity, it may give false positive cases, with patients without
nutritional risk being identified as at nutritional risk. On the other
hand, specificity refers to the proportion of “well nourished” cases
given by SGA and correctly classified as “without nutritional risk”
by the screening tools; a high specificity of a screening tool may
give false negative cases. The positive predictive value (PPV) is
defined as the proportion of patients classified as “at nutritional
risk” by the screening tools, who are correctly diagnosed by the
reference method, which was the SGA. Conversely, the negative
predictive value (NPV) is the proportion of patients identified as
“without nutritional risk” by the screening tools and also classified
as “well nourished” by the reference method. Statistical signifi-
cance was determined for p < 0.05.
3. Results
3.1. Patients’ characteristics
The overall studied cohort comprised 300 patients, whereas NRI
screening was calculated in a subset of 237 patients with available
plasma albumin values; patients’ characteristics are summarized in
Table 1. Among cancer patients, the most prevalent was colorectal
cancer (n ¼ 74, 25%); GI surgery was performed in 229 (76%)

Table 1
Patients’ characteristics.

Total (n ¼ 300) Cancer (n ¼ 139) Non cancer (n ¼ 161) NRI patients* (n ¼ 237) Cancer (n ¼ 117) Non cancer (n ¼ 120)
Men 132 (44%) 61 (20%) 71 (24%) 112 (47%) 56 (24%) 56 (24%)
Women 168 (56%) 78 (26%) 90 (30%) 125 (53%) 61 (26%) 64 (27%)
Age (years) 60  17 63  15 58  18 61  17 65  14 58  18
Age
65 years 134 (45%) 70 (23%) 64 (21%) 116 (49%) 65 (27%) 51 (22%)
Elective surgery 163 (54%) 84 (28%) 79 (26%) 121 (51%) 67 (28%) 54 (23%)
Non elective surgery 137 (46%) 55 (18%) 82 (27%) 116 (49%) 50 (21%) 66 (28%)
GI surgery 229 (76%) 103 (34%) 126 (42%) 193 (81%) 95 (40%) 98 (41%)
Other 71 (24%) 36 (12%) 35 (12%) 44 (19%) 22 (9%) 22 (9%)

Results expressed by number (percentage) of patients, mean  SD; NRI: Nutritional Risk Index; GI: gastrointestinal; *patients with available plasma albumin values in whom
NRI was calculated.
208 A.I. Almeida et al. / Clinical Nutrition 31 (2012) 206e211

patients, 113 (49%) of whom were admitted for non-elective admitted for GI surgery were more frequently identified as at
surgery. nutritional risk by both MUST and NRS-2002 (p < 0.001), Table 2.

3.2. Nutritional status 3.3.2. Nutritional risk index

Table 2 shows nutritional parameters results: nutritional status
by BMI, % of recent weight loss and SGA, as well as nutritional risk
results by NRS-2002, MUST and NRI, according to the major indi-
cation for surgery. After analyzing the detailed distribution of the
nutritional risk and status categories, we performed a frequency
analysis by comparing all methods as a first approach to the data.
The Chi-square test was used to explore the frequencies of the test
results (low risk vs at risk) in the various surgical groups, Table 2.
Patients admitted for cancer or GI surgery were more frequently
identified as undernourished or at nutritional risk by %weight loss,
SGA, NRS-2002 and MUST (p < 0.001). In every group, categories of
BMI nutritional status and NRI risk were either non-significant or
barely reached significance thus revealing a limited clinical value.
3.2.1. BMI and weight loss
Overall, BMI classified 18 (6%) patients as underweight
(<18.5 kg/m2), whereas 90 (30%) were overweight/obese. In the
previous six months before admission, a median involuntary
weight loss of 7 kg (range: 1e24) was accounted in 211 (70%)
patients; 53 (25%) of the latter were still classified as being over-
weight/obese.
3.2.2. Subjective global assessment
According to SGA, 193 (64%) patients were undernourished, 87
(29%) of whom had severe undernutrition. SGA undernutrition was
more frequently identified in cancer patients and in patients
admitted for GI surgery (p < 0.001), Table 2. Worth mentioning that
among the patients identified by SGA as undernourished, 24% were
classified by BMI as overweight or obese.
3.3. Nutritional risk
3.3.1. NRS-2002 and MUST
Both NRS-2002 and MUST identified 66% of patients at nutri-
tional risk. MUST further classified at high nutritional risk more
than half of patients (56%). Cancer patients as well as those
Overall, NRI classified 206/237 (87%) patients at nutritional risk,
126 (53%) of whom were identified as being at high risk, Tables 2
and 3. Of the 237 patients in whom NRI could be calculated, 16
(7%) were classified as underweight by BMI, whereas 61 (26%) were
overweight/obese. A recent %weight loss
5 of the usual body
weight was reported by 174/237 (73%) patients. In what concerns
SGA, 179/237 (76%) were classified as undernourished and 58 (24%)
were well nourished, Table 3.
3.4. Comparisons between methods
3.4.1. BMI or recent %weight loss vs SGA
In this section, we first undertook a frequency analysis by
comparing all methods, grouped into 2 categories to enable
comparisons, using the Chi-square test, Table 3. Results revealed
a significant association between %weight loss, NRS-2002, MUST
and NRI vs SGA (T ¼ 30.4; p < 0.001). Moreover, there was indeed
a remarkably high similarity between %weight loss, NRS-2002 and
MUST vs SGA (T ¼ 249; p < 0.001). Though significant, BMI vs SGA
showed the weakest results and the most discrepant in comparison
with the remaining methods.
To evaluate consistency and confirm the level of agreement
between methods’ results, a further in-depth analysis was under-
taken, by calculating the Kappa concordance coefficient and the
Spearman correlation, Table 4. The Kappa agreement coefficient
between BMI and SGA was low (kappa ¼ 0.068, p < 0.05);
conversely, an excellent Kappa was found between recent %weight
loss and SGA (kappa ¼ 0.871, p < 0.001), Table 4. By comparison
with SGA, only 18 (9%) patients were correctly identified by BMI as
undernourished, which leads to 175 (91%) patients with a BMI

18.5, falsely recognized as well nourished, overweight, or obese.
There were no false positive cases since all patients classified as
well nourished by SGA were similarly identified by BMI are well
nourished/overweight/obese.
Subsequently, to appraise the comparative value of nutritional
screening tools vs the reference SGA, sensitivity, specificity and
predictive values were mandatory, Table 5. When compared with

Table 2
Categories of nutritional risk and status in cancer surgery or other indications.

Cancer Elective surgery GI surgery

Yes No p Yes No p Yes No p

BMI

18.5 126 (42%) 156 (52%) <0.05 155 (52%) 127 (42%) NS 214 (71%) 68 (23%) NS
<18.5 13 (4%) 5 (2%) 8 (3%) 10 (3%) 15 (5%) 3 (1%)
% Weight loss
<5% 35 (12%) 80 (27%) <0.001 72 (24%) 43 (14%) <0.05 70 (23%) 45 (15%) <0.001

5% 104 (38%) 81 (27%) 91 (30%) 94 (31%) 159 (53%) 26 (9%)
SGA
Well nourished 28 (9%) 79 (26%) <0.001 71 (24%) 36 (12%) <0.01 63 (21%) 44 (15%) <0.001
Undernourished 111 (37%) 82 (27%) 92 (31%) 101 (34%) 166 (55%) 27 (9%)
NRS-2002
Low risk 27 (9%) 72 (24%) <0.001 69 (23%) 32 (11%) <0.01 58 (19%) 43 (14%) <0.001
Risk 111 (37%) 90 (30%) 94 (31%) 105 (35%) 173 (58%) 26 (9%)
MUST
Low risk 29 (10%) 72 (2%) <0.001 65 (22%) 36 (12%) <0.05 59 (20%) 42 (14%) <0.001
Risk 110 (37%) 89 (30%) 97 (32%) 102 (34%) 170 (58%) 29 (10%)
NRI
Low risk 12 (5%) 19 (8%) NS 21 (9%) 10 (4%) <0.05 23 (10%) 8 (3%) NS
Risk 105 (44%) 101 (43%) 100 (42%) 106 (45%) 170 (72%) 36 (15%)

Results expressed as number (percentage). BMI: Body Mass Index; SGA: Subjective Global Assessment; NRS-2002: Nutritional Risk Screening-2002; MUST: Malnutrition
Universal Screening Tool; NRI: Nutritional Risk Index (patients with plasma albumin); GI: gastrointestinal; NS: not significant. Statistical analysis was performed by T Test: Chi-
square frequency analysis of the test results (low risk vs risk) in cancer, GI or elective surgery.
 A.I. Almeida et al. / Clinical Nutrition 31 (2012) 206e211 209

Table 3
Comparison between SGA vs BMI, %weight loss, NRS-2002, MUST or NRI.

BMI* % Recent weight lossx NRS-2002x MUSTx NRIx

18.5 <18.5 <5%
5% Low risk Risk Low risk Risk Low risk Risk
SGA
Well nourished 107 (36%) 0 (0%) 102 (34%) 5 (2%) 94 (31%) 13 (4%) 99 (33%) 8 (3%) 20 (8%) 38 (16%)
Undernourished 175 (58%) 18 (6%) 13 (4%) 180 (60%) 7 (2%) 186 (62%) 4 (1%) 189 (63%) 11 (5%) 168 (71%)
Total 282 (94%) 18 (6%) 115 (38%) 185 (62%) 101 (34%) 199 (66%) 103 (34%) 197 (66%) 31 (13%) 206 (87%)

Results are expressed as number (percentage). BMI: Body Mass Index; NRS-2002: Nutritional Risk Screening 2002; MUST: Malnutrition Universal Screening Tool; NRI:
Nutritional Risk Index (patients with plasma albumin); SGA: Subjective Global Assessment. Frequency analysis by comparing all methods using Chi-square: SGA vs BMI, %
weight loss, NRS-2002, MUST, NRI; *p < 0.01; xp < 0.001.

SGA, BMI had a sensitivity of 0.43, NS (95% CI 0.33e0.47) and
a specificity of 0.39, NS (95% CI 0.35e0.42). BMI vs SGA had
a positive predictive value of 35% (NS) and a negative predictive
value of 31%, NS. Thus BMI had a weak capacity to detect patients at
risk of undernutrition, misclassifying a high number of patients
who were actually at risk.
When compared with SGA, %weight loss correctly classified as
undernourished 180 (93%) patients and 102 (95%) were correctly
classified as well nourished; %weight loss misclassified 13 (7%)
patients as well nourished and 5 (5%) as undernourished.
Percentage of weight loss had a sensitivity of 0.89, p < 0.001 (95% CI
0.85e0.92) and a specificity of 0.93, p < 0.001 (95% CI 0.87e0.96)
when compared with SGA; thus %weight loss had a positive
predictive value of 81% (p < 0.002) and a negative predictive value
of 89% (p < 0.001) vs SGA, Table 5. Indeed, %weight loss was highly
effective to detect undernourished patients.
at risk and 4 (2%) were also misclassified as without risk. The
sensitivity of MUST vs SGA was 0.85, p < 0.001 (95% CI 0.79e0.87)
and the specificity was 0.93, p < 0.001 (95% CI 0.87e0.95). MUST
had a 89% (p < 0.001) positive predictive value, and a negative
predictive value of 99% (p < 0.001), Table 5. Overall MUST showed
a high performance by comparison with the standard, and a strong
capacity to effectively detect patients at nutritional risk.
When compared with SGA, NRI correctly classified 168 (71%)
patients at nutritional risk, while 11 (5%) were misclassified as
without risk. On the other hand, NRI only correctly classified 20
(9%) patients as without risk, while 38 (17%) were falsely classified
as at risk. Sensitivity of NRI vs SGA was 0.29, NS (95% CI 0.26e0.33)
and the specificity was 0.27, NS (95% CI 0.23e0.29). NRI had
a positive predictive value of 24% (NS) and a negative predictive
value of 27%, NS, Table 5. Thus NRI had a weak capacity to detect
patients at risk of undernutrition.

3.4.2. Nutritional risk tools vs SGA
Both NRS-2002 and MUST showed a significant Kappa concor-
dance agreement with SGA (kappa ¼ 0.853 and kappa ¼ 0.912,
p < 0.001, respectively). Conversely, the Kappa between NRI and
SGA was low (kappa ¼ 0.336, p < 0.05), Table 4. When compared
with SGA, NRS-2002 correctly classified 186 (96%) patients as being
at nutritional risk and 94 (88%) were correctly classified as without
undernutrition risk. On the other hand, 7 (4%) patients were falsely
identified by NRS-2002 as being at risk, while 13 (12%) were also
misclassified as without risk. The sensitivity of NRS-2002 vs SGA
was 0.80, p < 0.001 (95% CI 0.76e0.84) and the specificity was 0.89,
p < 0.001 (95% CI 0.84e0.92). NRS-2002 when compared with SGA
had a 87% (p < 0.002) positive predictive value, and a negative
predictive value of 100% (p < 0.001). Overall NRS-2002 showed
a high performance by comparison with the standard, and a strong
capacity to effectively detect patients at nutritional risk.
When compared with SGA, MUST correctly classified 189 (98%)
patients as at nutritional risk and 99 (93%) as without risk.
Conversely, 8 (7%) patients were falsely classified by MUST as being
3.4.3. NRS-2002 or MUST vs %weight loss
Given the above results, recent %weight loss was tested to assess
its value per se. Both NRS-2002 and MUST showed a high Kappa
agreement coefficient with %weight loss (kappa ¼ 0.812, p < 0.001
and kappa ¼ 0.914, p < 0.001, respectively). Percentage weight loss
vs NRS-2002 correctly identified 179 (90%) patients at nutritional
risk and 95 (94%) without risk. Conversely, %weight loss falsely
identified 20 (10%) patients as without nutritional risk and 5 (6%) as
at risk; %weight loss had a sensitivity of 0.79, p < 0.002 (95% CI
0.76e0.81) and a specificity of 0.85, p < 0.001 (95% CI 0.81e0.89).
Furthermore, %weight loss vs NRS-2002 had a positive predictive
value of 84% (p < 0.001) and a negative predictive value of 87%
(p < 0.001).
When compared with MUST, %weight loss correctly identified
185 (94%) patients at nutritional risk; however, 12 (6%) patients
were falsely identified as without risk by % weight loss, while MUST
classified them as at nutritional risk. All patients recognized by
MUST as without nutritional risk were also identified by %weight
loss, thus there were no false positive cases. Percentage of weight

Table 4
Nutritional risk and status: Correlation and concordance between methods.
Table 5
BMI %Weight SGA NRS-2002 MUST NRI Nutritional risk tools, BMI, recent %weight loss vs SGA: sensitivity, specificity,
loss positive and negative predictive values.

BMI e 0.170* 0.188* 0.150* 0.183* 0.104 SGA (reference method)

(NS)
%Weight loss 0.065* e 0.873*** 0.817*** 0.917*** 0.316* Sensitivity Specificity PPV NPV
SGA 0.068* 0.871*** e 0.854*** 0.912*** 0.361* BMI 0.43 (0.33e0.47) 0.39 (0.35e0.42) 35% 31%
NRS-2002 0.052* 0.812*** 0.853*** e 0.822*** 0.369* % Weight loss* 0.89 (0.87e0.96) 0.93 (0.87e0.96) 81% 89%
MUST 0.065* 0.914*** 0.912*** 0.822*** e 0.369* NRS-2002* 0.8 (0.76e0.84) 0.89 (0.84e0.92) 87% 100%
NRI 0.022 0.329* 0.336* 0.352* 0.352* e MUST* 0.85 (0.79e0.87) 0.93 (0.87e0.95) 89% 99%
(NS) NRI 0.29 (0.26e0.33) 0.27 (0.23e0.29) 24% 27%

Spearman correlation coefficients are showed in bold, upper right; kappa coeffi-
cients in the lower left. BMI: Body Mass Index; SGA: Subjective Global Assessment;
NRS-2002: Nutritional Risk Screening-2002; MUST: Malnutrition Universal
Screening Tool; NRI: Nutritional Risk Index; NS: not significant, *p < 0.05, **p < 0.01,
***p < 0.001.
PPV: Positive predictive value; NPV: Negative predictive value; SGA: Subjective
Global Assessment; BMI: Body Mass Index; NRS-2002: Nutritional Risk Screening
2002; MUST: Malnutrition Universal Screening Tool; NRI: Nutritional Risk Index.
Sensitivity and specificity results expressed as percentage (95% CI); *statistically
significant.
210 A.I. Almeida et al. / Clinical Nutrition 31 (2012) 206e211
loss had a sensitivity of 0.87, p < 0.001 (95% CI 0.83e0.89) and
a specificity of 0.89, p < 0.001 (95% CI 0.86e0.95); %weight loss vs
MUST had a positive predictive value of 85% (p < 0.001) and
a negative predictive value of 89% (p < 0.001).
4. Discussion
Nutritional screening should be the first step to identify patients
at risk of nutritional depletion, for early referral to a complete
nutritional assessment and eventual intervention.17,21 In the
present study, different screening tools and nutritional parameters
were compared in order to establish a reliable and easy tool more
likely to become customary in daily hospital practice. Both MUST
and NRS-2002 showed a high performance to detect patients at
nutritional risk when compared with SGA. However, a weight loss

5% in the last 6 months also proved to be an effective screening
parameter to identify patients expected to be at risk of
undernutrition.
At hospital admission in surgical patients, our results showed
a remarkably similar performance of NRS-2002 and MUST to detect
patients at nutritional risk: both identified at risk 66% of patients
and SGA also identified an overlapping 64% of patients as being
undernourished. The prevalence of nutritional risk and/or under-
nutrition found in this cohort was higher than in other recent
studies,9,19,22 a discrepancy which may stem from different cohorts,
namely hospital departments and patients’ characteristics. Both
NRS-2002 and MUST were highly effective in the recognition of
patients at nutritional risk: their sensitivity, specificity, PPV and
NPV values were highly significant when compared with SGA. Our
results showed a higher Kappa concordant coeficients between
NRS-2002/MUST and SGA than other studies.9,19 Nonetheless,
similar values of sensitivity, specificity, PPV and NPV for NRS-2002
and MUST vs SGA as the standard method, were also found in both
studies.9,19
NRI vs SGA showed a higher percentage of patients at risk
though with a low sensitivity, specificity, PPV and NPV values. NRI
also failed to detect a higher risk ascertained by SGA to cancer
patients and those admitted for GI surgery. Although NRI was
devised for surgical patients, its estimation relies on serum albumin
concentration, a well known risk predictor in a broad sense rather
than a “marker” of undernutrition23,24. Although conscious of the
potential bias created by the lower number of patients assessed by
the NRI, our results do stress how undependable NRI may be for
nutritional screening in surgical patients.
BMI failed to identify patients at undernutrition; yet was the
only parameter able to identify overweight/obese patients.
Unintentional recent weight loss is present in NRS-2002, MUST
and SGA and is expressed by cut-offs that reflect practical bound-
aries between weight changes and their possible relation to
underlying conditions, which if undetected could worsen weight
loss and undernutrition. There is evidence that an unintentional
weight loss of 5%e10% may produce relevant changes in physio-
logical functions.25 In fact, our results showed a high performance
of
5%weight loss in detecting patients likely to be undernour-
ished; a high sensitivity, specificity, PPV and NPV were revealed
when compared with SGA. Indeed,
5%weight loss per se proved to
be a reliable nutritional parameter, with a high sensitivity of
0.79e0.87, specificity of 0.85e0.89, PPV of 84%e85% and NPV of
87%e89%, when compared with MUST and NRS-2002.
In fact, Boleo-Tomé et al. tested the ability of
5%weight loss to
detect undernutrition risk vs Patient Generated-Subjective Global
Assessment and MUST in ambulatory cancer patients; %weight loss
was shown to have a high sensitivity, specificity, PPV and NPV.26
Quick and easy %weight loss estimation may facilitate health
professionals’ adherence. In a recent study which aimed to foster
nutritional screening via teaching and training health professionals
on how to use MUST, weighing and calculating recent %weight loss
was unexpectedly adopted, incorporated and registered in patients’
records even by physicians.27
The early detection of patients at nutritional risk and their
referral to specialists for a comprehensive nutritional assessment
and individualized nutritional intervention is a major determinant
for quality of health care.28 Based in our results, both screening
methods NRS-2002 and MUST emerged as the most concordant,
valid and reliable tools to detect nutritional risk in surgical patients.
However,
5%weight loss in the previous 6 months also proved to
be as reliable and valid. Therefore, %weight loss estimation and
further registration should be the minimum and mandatory
parameter to be applied in routine practice, optimizing outcome-
driven nutritional management.
Conflict of interest
All authors hereby disclose any financial and personal rela-
tionships with other people or organization that could inappro-
priately influence our work.
Statement of authorship
AIA was responsible for data analysis, interpretation and writing
of the manuscript. M Correia was responsible for data collection. M
Camilo was responsible for writing and final review of the manu-
script as well as data interpretation. PR was responsible for writing
and final review of the manuscript, data interpretation and data
analysis.
Acknowledgments
This study was partially supported by a Grant from the “Fun-
dação para a Ciência e Tecnologia” (RUN 437).
References
1. Studley H. Percentage of weight loss: a basic indicator of surgical risk in
patients with chronic peptic ulcer. JAMA 1936;106:321e36.
2. Edington J, Boorman J, Durrant ER, Perkins A, Giffin CV, James R, et al. Preva-
lence of malnutrition on admission to four hospitals in England. Clin Nutr
2000;19(3):191e5.
3. Waitzberg D, Caiaffa W, Correia M. Hospital malnutrition: the Brazilian national
survey (IBRANUTRI): a study of 4000 patients. Nutrition 2001;17:573e80.
4. Valero M, Diez L, El Kadaoui N, Jiménez AE, Rodriguez H, León M. Are the tools
recommended by ASPEN and ESPEN comparable for assessing the nutritional
status? Nutr Hosp 2005;20:259e67.
5. Brunn L, Bosaeus I, Bergstad I, Nygaard K. Prevalence of malnutrition in surgical
patients: evaluation of nutritional support and documentation. Clin Nutr
1999;18(3):141e7.
6. Mourão F, Amado D, Ravasco P, Marques Vidal P, Camilo ME. Nutritional risk
and status assessment in surgical patients: a challenge amidst plenty. Nutr
Hosp 2004;19:83e8.
7. Correia M, Caiaffa W, Lázaro da Silva A, Waitzberg DL. Risk factors for
malnutrition in patients undergoing gastroenterological and hernia surgery: an
analysis of 374 patients. Nutr Hosp 2001;16(2):59e64.
8. Vidal A, Iglesias MJ, Pertega S, Ayúcar A, Vidal O. Prevalence of malnutrition in
medical and surgical wards of a university hospital. Nutr Hosp
2008;23:263e7.
9. Velasco C, Garcia E, Rodriguez V, Frias L, Garriga R, Alvarez J, et al. Comparison
of four nutritional screening tools to detect nutritional risk in hospitalized
patients: a multicentre study. Eur J Clin Nut 2011;65:269e74.
10. Norman K, Pichard C, Lochs H, Pirlich M. Prognostic impact of disease-related
malnutrition. Clin Nutr 2008;27:5e15.
11. Leonard-Jones J. A positive approach to nutrition as treatment. London: King’s
Fund Report; 2005.
12. O’Flynn J, Peake H, Hickson M, Foster D, Frost G. The prevalence of malnutrition
in hospitals can be reduced: results from three consecutive cross-sectional
studies. Clin Nutr 2005;24:1078e88.
13. Kondrup J, Allison SP, Elia M, Vellas B, Plauth M. ESPEN guidelines for nutri-
tional screening 2002. Clin Nutr 2003;22:415e21.
 A.I. Almeida et al. / Clinical Nutrition 31 (2012) 206e211
14. Elia M. Screening for malnutrition: a multidisciplinary responsibility. Development
and use of the Malnutrition Universal Screening Tool (MUST) for adults. Malnu-
trition Advisory Group, a Standing Committee of BAPEN; 2003.
15. Veterans Affair Total Parenteral Nutrition Cooperative Study Group. Perioper-
ative total parenteral nutrition in surgical patients. New Engl J Med
1991;324:525e32.
16. WHO. Consultation on Obesity. Geneva: World Health Organization; 2005.
17. Council of Europe e Committee of Ministers. Resolution ResAP (2003) 3 on food
and nutritional care in hospitals 2003.
18. Detsky AS, McLaughlin JR, Baker JP, Johnston N, Whittaker S, Mendelson RA, et al.
What is subjective global assessment of nutritional status? JPEN 1987;11:8e13.
19. Kyle U, Kossovsky MP, Karsegard VL, Pichard C. Comparison of tools for
nutritional assessment and screening at hospital admission: a population
study. Clin Nutr 2006;25:409e17.
20. Fleiss J. Statistical methods for rates and proportions. New York: John Wiley &
Sons; 1981.
21. ASPEN Board of Directors and The Clinical Guidelines Task Force. Guidelines for
the use of parenteral and enteral nutrition in adult and pediatric patients. JPEN
2002;26. 9SA-11SA.
211
22. Raslan M, Gonzalez MC, Dias MC, Nascimento M, Castro M, Marques P, et al.
Comparison of nutritional risk screening tools for predicting clinical outcomes
in hospitalized patients. Nutrition 2010;26:721e6.
23. Franch-Arcas G. The meaning of hypoalbuminaemia in clinical practice. Clin
Nutr 2001;20:265e9.
24. Fuhrman M. The albumin-nutrition connection: separating myth from fact.
Nutrition 2002;18:199e200.
25. Stratton RJ, Hackston A, Longmore D, Dixon R, Price S, Stroud M, et al.
Malnutrition in hospitalized outpatients and inpatients: prevalence, concur-
rent validity and ease of use of the "malnutrition universal screening tool"
(MUST) for adults. Brit J Nutr 2004;92:799e808.
26. Boléo-Tomé C, Chaves M, Monteiro Grillo I, Camilo ME, Ravasco P. Malnutrition
Universal Screening Tool (MUST): is validated in cancer! in press.
27. Boléo-Tomé C, Chaves M, Monteiro Grillo I, Camilo ME, Ravasco P. Teaching
nutrition integration: MUST screening in cancer. Oncologist
2011;16(2):239e45.
28. Joint Commission on Accreditation of Healthcare Organizations (JCAHO).
A new approach to JCAHO accreditation standards. Jt Comm Perspect
2002;22:4e5.